Datasets ¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

InterHand2.6M (ECCV'2020)

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

@InProceedings{Moon_2020_ECCV_InterHand2.6M,
author = {Moon, Gyeongsik and Yu, Shoou-I and Wen, He and Shiratori, Takaaki and Lee, Kyoung Mu},
title = {InterHand2.6M: A Dataset and Baseline for 3D Interacting Hand Pose Estimation from a Single RGB Image},
booktitle = {European Conference on Computer Vision (ECCV)},
year = {2020}
}

Results on InterHand2.6M val & test set

Train Set	Set	Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
Human_annot	val(M)	pose_resnet_50	256x256	0.973	0.828	5.15	ckpt	log
Human_annot	test(H)	pose_resnet_50	256x256	0.973	0.826	5.27	ckpt	log
Human_annot	test(M)	pose_resnet_50	256x256	0.975	0.841	4.90	ckpt	log
Human_annot	test(H+M)	pose_resnet_50	256x256	0.975	0.839	4.97	ckpt	log
Machine_annot	val(M)	pose_resnet_50	256x256	0.970	0.824	5.39	ckpt	log
Machine_annot	test(H)	pose_resnet_50	256x256	0.969	0.821	5.52	ckpt	log
Machine_annot	test(M)	pose_resnet_50	256x256	0.972	0.838	5.03	ckpt	log
Machine_annot	test(H+M)	pose_resnet_50	256x256	0.972	0.837	5.11	ckpt	log
All	val(M)	pose_resnet_50	256x256	0.977	0.840	4.66	ckpt	log
All	test(H)	pose_resnet_50	256x256	0.979	0.839	4.65	ckpt	log
All	test(M)	pose_resnet_50	256x256	0.979	0.838	4.42	ckpt	log
All	test(H+M)	pose_resnet_50	256x256	0.979	0.851	4.46	ckpt	log

Internet + Internet on Interhand3d¶

InterNet (ECCV'2020)

@InProceedings{Moon_2020_ECCV_InterHand2.6M,
author = {Moon, Gyeongsik and Yu, Shoou-I and Wen, He and Shiratori, Takaaki and Lee, Kyoung Mu},
title = {InterHand2.6M: A Dataset and Baseline for 3D Interacting Hand Pose Estimation from a Single RGB Image},
booktitle = {European Conference on Computer Vision (ECCV)},
year = {2020}
}

InterHand2.6M (ECCV'2020)

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

@InProceedings{Moon_2020_ECCV_InterHand2.6M,
author = {Moon, Gyeongsik and Yu, Shoou-I and Wen, He and Shiratori, Takaaki and Lee, Kyoung Mu},
title = {InterHand2.6M: A Dataset and Baseline for 3D Interacting Hand Pose Estimation from a Single RGB Image},
booktitle = {European Conference on Computer Vision (ECCV)},
year = {2020}
}

Results on InterHand2.6M val & test set

Train Set	Set	Arch	Input Size	MPJPE-single	MPJPE-interacting	MPJPE-all	MRRPE	APh	ckpt	log
All	test(H+M)	InterNet_resnet_50	256x256	9.47	13.40	11.59	29.28	0.99	ckpt	log
All	val(M)	InterNet_resnet_50	256x256	11.22	15.23	13.16	31.73	0.98	ckpt	log

MPII (CVPR’2014)¶

Deeppose + Resnet on Mpii¶

@inproceedings{toshev2014deeppose,
  title={Deeppose: Human pose estimation via deep neural networks},
  author={Toshev, Alexander and Szegedy, Christian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={1653--1660},
  year={2014}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

@inproceedings{andriluka14cvpr,
  author = {Mykhaylo Andriluka and Leonid Pishchulin and Peter Gehler and Schiele, Bernt},
  title = {2D Human Pose Estimation: New Benchmark and State of the Art Analysis},
  booktitle = {IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2014},
  month = {June}
}

Results on MPII val set

Arch	Input Size	Mean	Mean@0.1	ckpt	log
deeppose_resnet_50	256x256	0.825	0.174	ckpt	log
deeppose_resnet_101	256x256	0.841	0.193	ckpt	log
deeppose_resnet_152	256x256	0.850	0.198	ckpt	log

Deeppose + Resnet + Rle on Mpii¶

@inproceedings{toshev2014deeppose,
  title={Deeppose: Human pose estimation via deep neural networks},
  author={Toshev, Alexander and Szegedy, Christian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={1653--1660},
  year={2014}
}

RLE (ICCV'2021)

@inproceedings{li2021human,
  title={Human pose regression with residual log-likelihood estimation},
  author={Li, Jiefeng and Bian, Siyuan and Zeng, Ailing and Wang, Can and Pang, Bo and Liu, Wentao and Lu, Cewu},
  booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
  pages={11025--11034},
  year={2021}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

@inproceedings{andriluka14cvpr,
  author = {Mykhaylo Andriluka and Leonid Pishchulin and Peter Gehler and Schiele, Bernt},
  title = {2D Human Pose Estimation: New Benchmark and State of the Art Analysis},
  booktitle = {IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2014},
  month = {June}
}

Results on MPII val set

Arch	Input Size	Mean	Mean@0.1	ckpt	log
deeppose_resnet_50_rle	256x256	0.860	0.263	ckpt	log

Topdown Heatmap + Hrnet on Mpii¶

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

@inproceedings{andriluka14cvpr,
  author = {Mykhaylo Andriluka and Leonid Pishchulin and Peter Gehler and Schiele, Bernt},
  title = {2D Human Pose Estimation: New Benchmark and State of the Art Analysis},
  booktitle = {IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2014},
  month = {June}
}

Results on MPII val set

Arch	Input Size	Mean	Mean@0.1	ckpt	log
pose_hrnet_w32	256x256	0.900	0.334	ckpt	log
pose_hrnet_w48	256x256	0.901	0.337	ckpt	log

Topdown Heatmap + Mobilenetv2 on Mpii¶

@inproceedings{sandler2018mobilenetv2,
  title={Mobilenetv2: Inverted residuals and linear bottlenecks},
  author={Sandler, Mark and Howard, Andrew and Zhu, Menglong and Zhmoginov, Andrey and Chen, Liang-Chieh},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={4510--4520},
  year={2018}
}

@inproceedings{andriluka14cvpr,
  author = {Mykhaylo Andriluka and Leonid Pishchulin and Peter Gehler and Schiele, Bernt},
  title = {2D Human Pose Estimation: New Benchmark and State of the Art Analysis},
  booktitle = {IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2014},
  month = {June}
}

Results on MPII val set

Arch	Input Size	Mean	Mean@0.1	ckpt	log
pose_mobilenetv2	256x256	0.854	0.235	ckpt	log

Topdown Heatmap + Resnet on Mpii¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

@inproceedings{andriluka14cvpr,
  author = {Mykhaylo Andriluka and Leonid Pishchulin and Peter Gehler and Schiele, Bernt},
  title = {2D Human Pose Estimation: New Benchmark and State of the Art Analysis},
  booktitle = {IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2014},
  month = {June}
}

Results on MPII val set

Arch	Input Size	Mean	Mean@0.1	ckpt	log
pose_resnet_50	256x256	0.882	0.286	ckpt	log
pose_resnet_101	256x256	0.888	0.290	ckpt	log
pose_resnet_152	256x256	0.889	0.303	ckpt	log

Topdown Heatmap + CPM on Mpii¶

CPM (CVPR'2016)

@inproceedings{wei2016convolutional,
  title={Convolutional pose machines},
  author={Wei, Shih-En and Ramakrishna, Varun and Kanade, Takeo and Sheikh, Yaser},
  booktitle={Proceedings of the IEEE conference on Computer Vision and Pattern Recognition},
  pages={4724--4732},
  year={2016}
}

@inproceedings{andriluka14cvpr,
  author = {Mykhaylo Andriluka and Leonid Pishchulin and Peter Gehler and Schiele, Bernt},
  title = {2D Human Pose Estimation: New Benchmark and State of the Art Analysis},
  booktitle = {IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2014},
  month = {June}
}

Results on MPII val set

Arch	Input Size	Mean	Mean@0.1	ckpt	log
cpm	368x368	0.876	0.285	ckpt	log

Topdown Heatmap + Shufflenetv2 on Mpii¶

ShufflenetV2 (ECCV'2018)

@inproceedings{ma2018shufflenet,
  title={Shufflenet v2: Practical guidelines for efficient cnn architecture design},
  author={Ma, Ningning and Zhang, Xiangyu and Zheng, Hai-Tao and Sun, Jian},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={116--131},
  year={2018}
}

@inproceedings{andriluka14cvpr,
  author = {Mykhaylo Andriluka and Leonid Pishchulin and Peter Gehler and Schiele, Bernt},
  title = {2D Human Pose Estimation: New Benchmark and State of the Art Analysis},
  booktitle = {IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2014},
  month = {June}
}

Results on MPII val set

Arch	Input Size	Mean	Mean@0.1	ckpt	log
pose_shufflenetv2	256x256	0.828	0.205	ckpt	log

Topdown Heatmap + Litehrnet on Mpii¶

LiteHRNet (CVPR'2021)

@inproceedings{Yulitehrnet21,
  title={Lite-HRNet: A Lightweight High-Resolution Network},
  author={Yu, Changqian and Xiao, Bin and Gao, Changxin and Yuan, Lu and Zhang, Lei and Sang, Nong and Wang, Jingdong},
  booktitle={CVPR},
  year={2021}
}

@inproceedings{andriluka14cvpr,
  author = {Mykhaylo Andriluka and Leonid Pishchulin and Peter Gehler and Schiele, Bernt},
  title = {2D Human Pose Estimation: New Benchmark and State of the Art Analysis},
  booktitle = {IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2014},
  month = {June}
}

Results on MPII val set

Arch	Input Size	Mean	Mean@0.1	ckpt	log
LiteHRNet-18	256x256	0.859	0.260	ckpt	log
LiteHRNet-30	256x256	0.869	0.271	ckpt	log

Topdown Heatmap + Resnext on Mpii¶

ResNext (CVPR'2017)

@inproceedings{xie2017aggregated,
  title={Aggregated residual transformations for deep neural networks},
  author={Xie, Saining and Girshick, Ross and Doll{\'a}r, Piotr and Tu, Zhuowen and He, Kaiming},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={1492--1500},
  year={2017}
}

@inproceedings{andriluka14cvpr,
  author = {Mykhaylo Andriluka and Leonid Pishchulin and Peter Gehler and Schiele, Bernt},
  title = {2D Human Pose Estimation: New Benchmark and State of the Art Analysis},
  booktitle = {IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2014},
  month = {June}
}

Results on MPII val set

Arch	Input Size	Mean	Mean@0.1	ckpt	log
pose_resnext_152	256x256	0.887	0.294	ckpt	log

Topdown Heatmap + Shufflenetv1 on Mpii¶

ShufflenetV1 (CVPR'2018)

@inproceedings{zhang2018shufflenet,
  title={Shufflenet: An extremely efficient convolutional neural network for mobile devices},
  author={Zhang, Xiangyu and Zhou, Xinyu and Lin, Mengxiao and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={6848--6856},
  year={2018}
}

@inproceedings{andriluka14cvpr,
  author = {Mykhaylo Andriluka and Leonid Pishchulin and Peter Gehler and Schiele, Bernt},
  title = {2D Human Pose Estimation: New Benchmark and State of the Art Analysis},
  booktitle = {IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2014},
  month = {June}
}

Results on MPII val set

Arch	Input Size	Mean	Mean@0.1	ckpt	log
pose_shufflenetv1	256x256	0.823	0.195	ckpt	log

Topdown Heatmap + Hrnet + Dark on Mpii¶

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

@inproceedings{zhang2020distribution,
  title={Distribution-aware coordinate representation for human pose estimation},
  author={Zhang, Feng and Zhu, Xiatian and Dai, Hanbin and Ye, Mao and Zhu, Ce},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={7093--7102},
  year={2020}
}

@inproceedings{andriluka14cvpr,
  author = {Mykhaylo Andriluka and Leonid Pishchulin and Peter Gehler and Schiele, Bernt},
  title = {2D Human Pose Estimation: New Benchmark and State of the Art Analysis},
  booktitle = {IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2014},
  month = {June}
}

Results on MPII val set

Arch	Input Size	Mean	Mean@0.1	ckpt	log
pose_hrnet_w32_dark	256x256	0.904	0.354	ckpt	log
pose_hrnet_w48_dark	256x256	0.905	0.360	ckpt	log

Topdown Heatmap + Scnet on Mpii¶

@inproceedings{liu2020improving,
  title={Improving Convolutional Networks with Self-Calibrated Convolutions},
  author={Liu, Jiang-Jiang and Hou, Qibin and Cheng, Ming-Ming and Wang, Changhu and Feng, Jiashi},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={10096--10105},
  year={2020}
}

@inproceedings{andriluka14cvpr,
  author = {Mykhaylo Andriluka and Leonid Pishchulin and Peter Gehler and Schiele, Bernt},
  title = {2D Human Pose Estimation: New Benchmark and State of the Art Analysis},
  booktitle = {IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2014},
  month = {June}
}

Results on MPII val set

Arch	Input Size	Mean	Mean@0.1	ckpt	log
pose_scnet_50	256x256	0.888	0.290	ckpt	log
pose_scnet_101	256x256	0.886	0.293	ckpt	log

Topdown Heatmap + Seresnet on Mpii¶

SEResNet (CVPR'2018)

@inproceedings{hu2018squeeze,
  title={Squeeze-and-excitation networks},
  author={Hu, Jie and Shen, Li and Sun, Gang},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={7132--7141},
  year={2018}
}

@inproceedings{andriluka14cvpr,
  author = {Mykhaylo Andriluka and Leonid Pishchulin and Peter Gehler and Schiele, Bernt},
  title = {2D Human Pose Estimation: New Benchmark and State of the Art Analysis},
  booktitle = {IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2014},
  month = {June}
}

Results on MPII val set

Arch	Input Size	Mean	Mean@0.1	ckpt	log
pose_seresnet_50	256x256	0.884	0.292	ckpt	log
pose_seresnet_101	256x256	0.884	0.295	ckpt	log
pose_seresnet_152*	256x256	0.884	0.287	ckpt	log

Note that * means without imagenet pre-training.

Topdown Heatmap + Resnetv1d on Mpii¶

ResNetV1D (CVPR'2019)

@inproceedings{he2019bag,
  title={Bag of tricks for image classification with convolutional neural networks},
  author={He, Tong and Zhang, Zhi and Zhang, Hang and Zhang, Zhongyue and Xie, Junyuan and Li, Mu},
  booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
  pages={558--567},
  year={2019}
}

@inproceedings{andriluka14cvpr,
  author = {Mykhaylo Andriluka and Leonid Pishchulin and Peter Gehler and Schiele, Bernt},
  title = {2D Human Pose Estimation: New Benchmark and State of the Art Analysis},
  booktitle = {IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2014},
  month = {June}
}

Results on MPII val set

Arch	Input Size	Mean	Mean@0.1	ckpt	log
pose_resnetv1d_50	256x256	0.881	0.290	ckpt	log
pose_resnetv1d_101	256x256	0.883	0.295	ckpt	log
pose_resnetv1d_152	256x256	0.888	0.300	ckpt	log

Topdown Heatmap + Hourglass on Mpii¶

@inproceedings{newell2016stacked,
  title={Stacked hourglass networks for human pose estimation},
  author={Newell, Alejandro and Yang, Kaiyu and Deng, Jia},
  booktitle={European conference on computer vision},
  pages={483--499},
  year={2016},
  organization={Springer}
}

@inproceedings{andriluka14cvpr,
  author = {Mykhaylo Andriluka and Leonid Pishchulin and Peter Gehler and Schiele, Bernt},
  title = {2D Human Pose Estimation: New Benchmark and State of the Art Analysis},
  booktitle = {IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2014},
  month = {June}
}

Results on MPII val set

Arch	Input Size	Mean	Mean@0.1	ckpt	log
pose_hourglass_52	256x256	0.889	0.317	ckpt	log
pose_hourglass_52	384x384	0.894	0.366	ckpt	log

NVGesture (CVPR’2016)¶

Mtut + I3d on Nvgesture¶

MTUT (CVPR'2019)

@InProceedings{Abavisani_2019_CVPR,
  author = {Abavisani, Mahdi and Joze, Hamid Reza Vaezi and Patel, Vishal M.},
  title = {Improving the Performance of Unimodal Dynamic Hand-Gesture Recognition With Multimodal Training},
  booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  month = {June},
  year = {2019}
}

I3D (CVPR'2017)

@InProceedings{Carreira_2017_CVPR,
  author = {Carreira, Joao and Zisserman, Andrew},
  title = {Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset},
  booktitle = {Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  month = {July},
  year = {2017}
}

NVGesture (CVPR'2016)

@InProceedings{Molchanov_2016_CVPR,
  author = {Molchanov, Pavlo and Yang, Xiaodong and Gupta, Shalini and Kim, Kihwan and Tyree, Stephen and Kautz, Jan},
  title = {Online Detection and Classification of Dynamic Hand Gestures With Recurrent 3D Convolutional Neural Network},
  booktitle = {Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  month = {June},
  year = {2016}
}

Results on NVGesture test set

Arch	Input Size	fps	bbox	AP_rgb	AP_depth	ckpt	log
I3D+MTUT^*	112x112	15	$\surd$	0.725	0.730	ckpt	log
I3D+MTUT	224x224	30	$\surd$	0.782	0.811	ckpt	log
I3D+MTUT	224x224	30	$\times$	0.739	0.809	ckpt	log

^*: MTUT supports multi-modal training and uni-modal testing. Model trained with this config can be used to recognize gestures in rgb videos with inference config.

MacaquePose (bioRxiv’2020)¶

Topdown Heatmap + Resnet on Macaque¶

MacaquePose (bioRxiv'2020)

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

@article{labuguen2020macaquepose,
  title={MacaquePose: A novel ‘in the wild’macaque monkey pose dataset for markerless motion capture},
  author={Labuguen, Rollyn and Matsumoto, Jumpei and Negrete, Salvador and Nishimaru, Hiroshi and Nishijo, Hisao and Takada, Masahiko and Go, Yasuhiro and Inoue, Ken-ichi and Shibata, Tomohiro},
  journal={bioRxiv},
  year={2020},
  publisher={Cold Spring Harbor Laboratory}
}

Results on MacaquePose with ground-truth detection bounding boxes

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_resnet_50	256x192	0.799	0.952	0.919	0.837	0.964	ckpt	log
pose_resnet_101	256x192	0.790	0.953	0.908	0.828	0.967	ckpt	log
pose_resnet_152	256x192	0.794	0.951	0.915	0.834	0.968	ckpt	log

Topdown Heatmap + Hrnet on Macaque¶

MacaquePose (bioRxiv'2020)

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

@article{labuguen2020macaquepose,
  title={MacaquePose: A novel ‘in the wild’macaque monkey pose dataset for markerless motion capture},
  author={Labuguen, Rollyn and Matsumoto, Jumpei and Negrete, Salvador and Nishimaru, Hiroshi and Nishijo, Hisao and Takada, Masahiko and Go, Yasuhiro and Inoue, Ken-ichi and Shibata, Tomohiro},
  journal={bioRxiv},
  year={2020},
  publisher={Cold Spring Harbor Laboratory}
}

Results on MacaquePose with ground-truth detection bounding boxes

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_hrnet_w32	256x192	0.814	0.953	0.918	0.851	0.969	ckpt	log
pose_hrnet_w48	256x192	0.818	0.963	0.917	0.855	0.971	ckpt	log

ATRW (ACM MM’2020)¶

Topdown Heatmap + Hrnet on Atrw¶

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

ATRW (ACM MM'2020)

@inproceedings{li2020atrw,
  title={ATRW: A Benchmark for Amur Tiger Re-identification in the Wild},
  author={Li, Shuyuan and Li, Jianguo and Tang, Hanlin and Qian, Rui and Lin, Weiyao},
  booktitle={Proceedings of the 28th ACM International Conference on Multimedia},
  pages={2590--2598},
  year={2020}
}

Results on ATRW validation set

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_hrnet_w32	256x256	0.912	0.973	0.959	0.938	0.985	ckpt	log
pose_hrnet_w48	256x256	0.911	0.972	0.946	0.937	0.985	ckpt	log

Topdown Heatmap + Resnet on Atrw¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

ATRW (ACM MM'2020)

@inproceedings{li2020atrw,
  title={ATRW: A Benchmark for Amur Tiger Re-identification in the Wild},
  author={Li, Shuyuan and Li, Jianguo and Tang, Hanlin and Qian, Rui and Lin, Weiyao},
  booktitle={Proceedings of the 28th ACM International Conference on Multimedia},
  pages={2590--2598},
  year={2020}
}

Results on ATRW validation set

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_resnet_50	256x256	0.900	0.973	0.932	0.929	0.985	ckpt	log
pose_resnet_101	256x256	0.898	0.973	0.936	0.927	0.985	ckpt	log
pose_resnet_152	256x256	0.896	0.973	0.931	0.927	0.985	ckpt	log

Animal-Pose (ICCV’2019)¶

Topdown Heatmap + Hrnet on Animalpose¶

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

Animal-Pose (ICCV'2019)

@InProceedings{Cao_2019_ICCV,
    author = {Cao, Jinkun and Tang, Hongyang and Fang, Hao-Shu and Shen, Xiaoyong and Lu, Cewu and Tai, Yu-Wing},
    title = {Cross-Domain Adaptation for Animal Pose Estimation},
    booktitle = {The IEEE International Conference on Computer Vision (ICCV)},
    month = {October},
    year = {2019}
}

Results on AnimalPose validation set (1117 instances)

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_hrnet_w32	256x256	0.736	0.959	0.832	0.775	0.966	ckpt	log
pose_hrnet_w48	256x256	0.737	0.959	0.823	0.778	0.962	ckpt	log

Topdown Heatmap + Resnet on Animalpose¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

Animal-Pose (ICCV'2019)

@InProceedings{Cao_2019_ICCV,
    author = {Cao, Jinkun and Tang, Hongyang and Fang, Hao-Shu and Shen, Xiaoyong and Lu, Cewu and Tai, Yu-Wing},
    title = {Cross-Domain Adaptation for Animal Pose Estimation},
    booktitle = {The IEEE International Conference on Computer Vision (ICCV)},
    month = {October},
    year = {2019}
}

Results on AnimalPose validation set (1117 instances)

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_resnet_50	256x256	0.688	0.945	0.772	0.733	0.952	ckpt	log
pose_resnet_101	256x256	0.696	0.948	0.785	0.737	0.954	ckpt	log
pose_resnet_152	256x256	0.709	0.948	0.797	0.749	0.951	ckpt	log

Human3.6M (TPAMI’2014)¶

Topdown Heatmap + Hrnet on H36m¶

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

SimpleBaseline3D (ICCV'2017)

@article{h36m_pami,
  author = {Ionescu, Catalin and Papava, Dragos and Olaru, Vlad and Sminchisescu,  Cristian},
  title = {Human3.6M: Large Scale Datasets and Predictive Methods for 3D Human Sensing in Natural Environments},
  journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
  publisher = {IEEE Computer Society},
  volume = {36},
  number = {7},
  pages = {1325-1339},
  month = {jul},
  year = {2014}
}

Results on Human3.6M test set with ground truth 2D detections

Arch	Input Size	EPE	PCK	ckpt	log
pose_hrnet_w32	256x256	9.43	0.911	ckpt	log
pose_hrnet_w48	256x256	7.36	0.932	ckpt	log

Pose Lift + Simplebaseline3d on H36m¶

@inproceedings{martinez_2017_3dbaseline,
  title={A simple yet effective baseline for 3d human pose estimation},
  author={Martinez, Julieta and Hossain, Rayat and Romero, Javier and Little, James J.},
  booktitle={ICCV},
  year={2017}
}

@article{h36m_pami,
  author = {Ionescu, Catalin and Papava, Dragos and Olaru, Vlad and Sminchisescu,  Cristian},
  title = {Human3.6M: Large Scale Datasets and Predictive Methods for 3D Human Sensing in Natural Environments},
  journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
  publisher = {IEEE Computer Society},
  volume = {36},
  number = {7},
  pages = {1325-1339},
  month = {jul},
  year = {2014}
}

Results on Human3.6M dataset with ground truth 2D detections

Arch	MPJPE	P-MPJPE	ckpt	log
simple_baseline_3d_tcn¹	43.4	34.3	ckpt	log

¹ Differing from the original paper, we didn’t apply the max-norm constraint because we found this led to a better convergence and performance.

Video Pose Lift + Videopose3d on H36m¶

VideoPose3D (CVPR'2019)

@inproceedings{pavllo20193d,
  title={3d human pose estimation in video with temporal convolutions and semi-supervised training},
  author={Pavllo, Dario and Feichtenhofer, Christoph and Grangier, David and Auli, Michael},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={7753--7762},
  year={2019}
}

@article{h36m_pami,
  author = {Ionescu, Catalin and Papava, Dragos and Olaru, Vlad and Sminchisescu,  Cristian},
  title = {Human3.6M: Large Scale Datasets and Predictive Methods for 3D Human Sensing in Natural Environments},
  journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
  publisher = {IEEE Computer Society},
  volume = {36},
  number = {7},
  pages = {1325-1339},
  month = {jul},
  year = {2014}
}

Results on Human3.6M dataset with ground truth 2D detections, supervised training

Arch	Receptive Field	MPJPE	P-MPJPE	ckpt	log
VideoPose3D	27	40.0	30.1	ckpt	log
VideoPose3D	81	38.9	29.2	ckpt	log
VideoPose3D	243	37.6	28.3	ckpt	log

Results on Human3.6M dataset with CPN 2D detections¹, supervised training

Arch	Receptive Field	MPJPE	P-MPJPE	ckpt	log
VideoPose3D	1	52.9	41.3	ckpt	log
VideoPose3D	243	47.9	38.0	ckpt	log

Results on Human3.6M dataset with ground truth 2D detections, semi-supervised training

Training Data	Arch	Receptive Field	MPJPE	P-MPJPE	N-MPJPE	ckpt	log
10% S1	VideoPose3D	27	58.1	42.8	54.7	ckpt	log

Results on Human3.6M dataset with CPN 2D detections¹, semi-supervised training

Training Data	Arch	Receptive Field	MPJPE	P-MPJPE	N-MPJPE	ckpt	log
10% S1	VideoPose3D	27	67.4	50.1	63.2	ckpt	log

¹ CPN 2D detections are provided by official repo. The reformatted version used in this repository can be downloaded from train_detection and test_detection.

HMR + Resnet on Mixed¶

HMR (CVPR'2018)

@inProceedings{kanazawaHMR18,
  title={End-to-end Recovery of Human Shape and Pose},
  author = {Angjoo Kanazawa
  and Michael J. Black
  and David W. Jacobs
  and Jitendra Malik},
  booktitle={Computer Vision and Pattern Recognition (CVPR)},
  year={2018}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

@article{h36m_pami,
  author = {Ionescu, Catalin and Papava, Dragos and Olaru, Vlad and Sminchisescu,  Cristian},
  title = {Human3.6M: Large Scale Datasets and Predictive Methods for 3D Human Sensing in Natural Environments},
  journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
  publisher = {IEEE Computer Society},
  volume = {36},
  number = {7},
  pages = {1325-1339},
  month = {jul},
  year = {2014}
}

Results on Human3.6M with ground-truth bounding box having MPJPE-PA of 52.60 mm on Protocol2

Arch	Input Size	MPJPE (P1)	MPJPE-PA (P1)	MPJPE (P2)	MPJPE-PA (P2)	ckpt	log
hmr_resnet_50	224x224	80.75	55.08	80.35	52.60	ckpt	log

OneHand10K (TCSVT’2019)¶

Deeppose + Resnet on Onehand10k¶

@inproceedings{toshev2014deeppose,
  title={Deeppose: Human pose estimation via deep neural networks},
  author={Toshev, Alexander and Szegedy, Christian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={1653--1660},
  year={2014}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

@article{wang2018mask,
  title={Mask-pose cascaded cnn for 2d hand pose estimation from single color image},
  author={Wang, Yangang and Peng, Cong and Liu, Yebin},
  journal={IEEE Transactions on Circuits and Systems for Video Technology},
  volume={29},
  number={11},
  pages={3258--3268},
  year={2018},
  publisher={IEEE}
}

Results on OneHand10K val set

Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
deeppose_resnet_50	256x256	0.990	0.486	34.28	ckpt	log

Topdown Heatmap + Hrnetv2 + Dark on Onehand10k¶

@article{WangSCJDZLMTWLX19,
  title={Deep High-Resolution Representation Learning for Visual Recognition},
  author={Jingdong Wang and Ke Sun and Tianheng Cheng and
          Borui Jiang and Chaorui Deng and Yang Zhao and Dong Liu and Yadong Mu and
          Mingkui Tan and Xinggang Wang and Wenyu Liu and Bin Xiao},
  journal={TPAMI},
  year={2019}
}

@inproceedings{zhang2020distribution,
  title={Distribution-aware coordinate representation for human pose estimation},
  author={Zhang, Feng and Zhu, Xiatian and Dai, Hanbin and Ye, Mao and Zhu, Ce},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={7093--7102},
  year={2020}
}

@article{wang2018mask,
  title={Mask-pose cascaded cnn for 2d hand pose estimation from single color image},
  author={Wang, Yangang and Peng, Cong and Liu, Yebin},
  journal={IEEE Transactions on Circuits and Systems for Video Technology},
  volume={29},
  number={11},
  pages={3258--3268},
  year={2018},
  publisher={IEEE}
}

Results on OneHand10K val set

Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
pose_hrnetv2_w18_dark	256x256	0.990	0.573	23.84	ckpt	log

Topdown Heatmap + Resnet on Onehand10k¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

@article{wang2018mask,
  title={Mask-pose cascaded cnn for 2d hand pose estimation from single color image},
  author={Wang, Yangang and Peng, Cong and Liu, Yebin},
  journal={IEEE Transactions on Circuits and Systems for Video Technology},
  volume={29},
  number={11},
  pages={3258--3268},
  year={2018},
  publisher={IEEE}
}

Results on OneHand10K val set

Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
pose_resnet_50	256x256	0.989	0.555	25.19	ckpt	log

Topdown Heatmap + Hrnetv2 on Onehand10k¶

@article{WangSCJDZLMTWLX19,
  title={Deep High-Resolution Representation Learning for Visual Recognition},
  author={Jingdong Wang and Ke Sun and Tianheng Cheng and
          Borui Jiang and Chaorui Deng and Yang Zhao and Dong Liu and Yadong Mu and
          Mingkui Tan and Xinggang Wang and Wenyu Liu and Bin Xiao},
  journal={TPAMI},
  year={2019}
}

@article{wang2018mask,
  title={Mask-pose cascaded cnn for 2d hand pose estimation from single color image},
  author={Wang, Yangang and Peng, Cong and Liu, Yebin},
  journal={IEEE Transactions on Circuits and Systems for Video Technology},
  volume={29},
  number={11},
  pages={3258--3268},
  year={2018},
  publisher={IEEE}
}

Results on OneHand10K val set

Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
pose_hrnetv2_w18	256x256	0.990	0.568	24.16	ckpt	log

Topdown Heatmap + Mobilenetv2 on Onehand10k¶

@inproceedings{sandler2018mobilenetv2,
  title={Mobilenetv2: Inverted residuals and linear bottlenecks},
  author={Sandler, Mark and Howard, Andrew and Zhu, Menglong and Zhmoginov, Andrey and Chen, Liang-Chieh},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={4510--4520},
  year={2018}
}

@article{wang2018mask,
  title={Mask-pose cascaded cnn for 2d hand pose estimation from single color image},
  author={Wang, Yangang and Peng, Cong and Liu, Yebin},
  journal={IEEE Transactions on Circuits and Systems for Video Technology},
  volume={29},
  number={11},
  pages={3258--3268},
  year={2018},
  publisher={IEEE}
}

Results on OneHand10K val set

Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
pose_mobilenet_v2	256x256	0.986	0.537	28.60	ckpt	log

Topdown Heatmap + Hrnetv2 + Udp on Onehand10k¶

@article{WangSCJDZLMTWLX19,
  title={Deep High-Resolution Representation Learning for Visual Recognition},
  author={Jingdong Wang and Ke Sun and Tianheng Cheng and
          Borui Jiang and Chaorui Deng and Yang Zhao and Dong Liu and Yadong Mu and
          Mingkui Tan and Xinggang Wang and Wenyu Liu and Bin Xiao},
  journal={TPAMI},
  year={2019}
}

@InProceedings{Huang_2020_CVPR,
  author = {Huang, Junjie and Zhu, Zheng and Guo, Feng and Huang, Guan},
  title = {The Devil Is in the Details: Delving Into Unbiased Data Processing for Human Pose Estimation},
  booktitle = {The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  month = {June},
  year = {2020}
}

@article{wang2018mask,
  title={Mask-pose cascaded cnn for 2d hand pose estimation from single color image},
  author={Wang, Yangang and Peng, Cong and Liu, Yebin},
  journal={IEEE Transactions on Circuits and Systems for Video Technology},
  volume={29},
  number={11},
  pages={3258--3268},
  year={2018},
  publisher={IEEE}
}

Results on OneHand10K val set

Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
pose_hrnetv2_w18_udp	256x256	0.990	0.572	23.87	ckpt	log

JHMDB (ICCV’2013)¶

Topdown Heatmap + CPM on JHMDB¶

CPM (CVPR'2016)

@inproceedings{wei2016convolutional,
  title={Convolutional pose machines},
  author={Wei, Shih-En and Ramakrishna, Varun and Kanade, Takeo and Sheikh, Yaser},
  booktitle={Proceedings of the IEEE conference on Computer Vision and Pattern Recognition},
  pages={4724--4732},
  year={2016}
}

JHMDB (ICCV'2013)

@inproceedings{Jhuang:ICCV:2013,
  title = {Towards understanding action recognition},
  author = {H. Jhuang and J. Gall and S. Zuffi and C. Schmid and M. J. Black},
  booktitle = {International Conf. on Computer Vision (ICCV)},
  month = Dec,
  pages = {3192-3199},
  year = {2013}
}

Results on Sub-JHMDB dataset

The models are pre-trained on MPII dataset only. NO test-time augmentation (multi-scale /rotation testing) is used.

Normalized by Person Size

Split	Arch	Input Size	Head	Sho	Elb	Wri	Hip	Knee	Ank	Mean	ckpt	log
Sub1	cpm	368x368	96.1	91.9	81.0	78.9	96.6	90.8	87.3	89.5	ckpt	log
Sub2	cpm	368x368	98.1	93.6	77.1	70.9	94.0	89.1	84.7	87.4	ckpt	log
Sub3	cpm	368x368	97.9	94.9	87.3	84.0	98.6	94.4	86.2	92.4	ckpt	log
Average	cpm	368x368	97.4	93.5	81.5	77.9	96.4	91.4	86.1	89.8	-	-

Normalized by Torso Size

Split	Arch	Input Size	Head	Sho	Elb	Wri	Hip	Knee	Ank	Mean	ckpt	log
Sub1	cpm	368x368	89.0	63.0	54.0	54.9	68.2	63.1	61.2	66.0	ckpt	log
Sub2	cpm	368x368	90.3	57.9	46.8	44.3	60.8	58.2	62.4	61.1	ckpt	log
Sub3	cpm	368x368	91.0	72.6	59.9	54.0	73.2	68.5	65.8	70.3	ckpt	log
Average	cpm	368x368	90.1	64.5	53.6	51.1	67.4	63.3	63.1	65.7	-	-

Topdown Heatmap + Resnet on JHMDB¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

JHMDB (ICCV'2013)

@inproceedings{Jhuang:ICCV:2013,
  title = {Towards understanding action recognition},
  author = {H. Jhuang and J. Gall and S. Zuffi and C. Schmid and M. J. Black},
  booktitle = {International Conf. on Computer Vision (ICCV)},
  month = Dec,
  pages = {3192-3199},
  year = {2013}
}

Results on Sub-JHMDB dataset

The models are pre-trained on MPII dataset only. NO test-time augmentation (multi-scale /rotation testing) is used.

Normalized by Person Size

Split	Arch	Input Size	Head	Sho	Elb	Wri	Hip	Knee	Ank	Mean	ckpt	log
Sub1	pose_resnet_50	256x256	99.1	98.0	93.8	91.3	99.4	96.5	92.8	96.1	ckpt	log
Sub2	pose_resnet_50	256x256	99.3	97.1	90.6	87.0	98.9	96.3	94.1	95.0	ckpt	log
Sub3	pose_resnet_50	256x256	99.0	97.9	94.0	91.6	99.7	98.0	94.7	96.7	ckpt	log
Average	pose_resnet_50	256x256	99.2	97.7	92.8	90.0	99.3	96.9	93.9	96.0	-	-
Sub1	pose_resnet_50 (2 Deconv.)	256x256	99.1	98.5	94.6	92.0	99.4	94.6	92.5	96.1	ckpt	log
Sub2	pose_resnet_50 (2 Deconv.)	256x256	99.3	97.8	91.0	87.0	99.1	96.5	93.8	95.2	ckpt	log
Sub3	pose_resnet_50 (2 Deconv.)	256x256	98.8	98.4	94.3	92.1	99.8	97.5	93.8	96.7	ckpt	log
Average	pose_resnet_50 (2 Deconv.)	256x256	99.1	98.2	93.3	90.4	99.4	96.2	93.4	96.0	-	-

Normalized by Torso Size

Split	Arch	Input Size	Head	Sho	Elb	Wri	Hip	Knee	Ank	Mean	ckpt	log
Sub1	pose_resnet_50	256x256	93.3	83.2	74.4	72.7	85.0	81.2	78.9	81.9	ckpt	log
Sub2	pose_resnet_50	256x256	94.1	74.9	64.5	62.5	77.9	71.9	78.6	75.5	ckpt	log
Sub3	pose_resnet_50	256x256	97.0	82.2	74.9	70.7	84.7	83.7	84.2	82.9	ckpt	log
Average	pose_resnet_50	256x256	94.8	80.1	71.3	68.6	82.5	78.9	80.6	80.1	-	-
Sub1	pose_resnet_50 (2 Deconv.)	256x256	92.4	80.6	73.2	70.5	82.3	75.4	75.0	79.2	ckpt	log
Sub2	pose_resnet_50 (2 Deconv.)	256x256	93.4	73.6	63.8	60.5	75.1	68.4	75.5	73.7	ckpt	log
Sub3	pose_resnet_50 (2 Deconv.)	256x256	96.1	81.2	72.6	67.9	83.6	80.9	81.5	81.2	ckpt	log
Average	pose_resnet_50 (2 Deconv.)	256x256	94.0	78.5	69.9	66.3	80.3	74.9	77.3	78.0	-	-

CMU Panoptic (ICCV’2015)¶

Voxelpose + Voxelpose + Prn64x64x64 + Cpn80x80x20 + Panoptic on Panoptic¶

VoxelPose (ECCV'2020)

@inproceedings{tumultipose,
  title={VoxelPose: Towards Multi-Camera 3D Human Pose Estimation in Wild Environment},
  author={Tu, Hanyue and Wang, Chunyu and Zeng, Wenjun},
  booktitle={ECCV},
  year={2020}
}

CMU Panoptic (ICCV'2015)

@Article = {joo_iccv_2015,
author = {Hanbyul Joo, Hao Liu, Lei Tan, Lin Gui, Bart Nabbe, Iain Matthews, Takeo Kanade, Shohei Nobuhara, and Yaser Sheikh},
title = {Panoptic Studio: A Massively Multiview System for Social Motion Capture},
booktitle = {ICCV},
year = {2015}
}

Results on CMU Panoptic dataset.

Arch	mAP	mAR	MPJPE	Recall@500mm	ckpt	log
prn64_cpn80_res50	97.31	97.99	17.57	99.85	ckpt	log

CMU Panoptic HandDB (CVPR’2017)¶

Deeppose + Resnet on Panoptic2d¶

@inproceedings{toshev2014deeppose,
  title={Deeppose: Human pose estimation via deep neural networks},
  author={Toshev, Alexander and Szegedy, Christian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={1653--1660},
  year={2014}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

@inproceedings{simon2017hand,
  title={Hand keypoint detection in single images using multiview bootstrapping},
  author={Simon, Tomas and Joo, Hanbyul and Matthews, Iain and Sheikh, Yaser},
  booktitle={Proceedings of the IEEE conference on Computer Vision and Pattern Recognition},
  pages={1145--1153},
  year={2017}
}

Results on CMU Panoptic (MPII+NZSL val set)

Arch	Input Size	PCKh@0.7	AUC	EPE	ckpt	log
deeppose_resnet_50	256x256	0.999	0.686	9.36	ckpt	log

Topdown Heatmap + Mobilenetv2 on Panoptic2d¶

@inproceedings{sandler2018mobilenetv2,
  title={Mobilenetv2: Inverted residuals and linear bottlenecks},
  author={Sandler, Mark and Howard, Andrew and Zhu, Menglong and Zhmoginov, Andrey and Chen, Liang-Chieh},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={4510--4520},
  year={2018}
}

@inproceedings{simon2017hand,
  title={Hand keypoint detection in single images using multiview bootstrapping},
  author={Simon, Tomas and Joo, Hanbyul and Matthews, Iain and Sheikh, Yaser},
  booktitle={Proceedings of the IEEE conference on Computer Vision and Pattern Recognition},
  pages={1145--1153},
  year={2017}
}

Results on CMU Panoptic (MPII+NZSL val set)

Arch	Input Size	PCKh@0.7	AUC	EPE	ckpt	log
pose_mobilenet_v2	256x256	0.998	0.694	9.70	ckpt	log

Topdown Heatmap + Resnet on Panoptic2d¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

@inproceedings{simon2017hand,
  title={Hand keypoint detection in single images using multiview bootstrapping},
  author={Simon, Tomas and Joo, Hanbyul and Matthews, Iain and Sheikh, Yaser},
  booktitle={Proceedings of the IEEE conference on Computer Vision and Pattern Recognition},
  pages={1145--1153},
  year={2017}
}

Results on CMU Panoptic (MPII+NZSL val set)

Arch	Input Size	PCKh@0.7	AUC	EPE	ckpt	log
pose_resnet_50	256x256	0.999	0.713	9.00	ckpt	log

Topdown Heatmap + Hrnetv2 + Dark on Panoptic2d¶

@article{WangSCJDZLMTWLX19,
  title={Deep High-Resolution Representation Learning for Visual Recognition},
  author={Jingdong Wang and Ke Sun and Tianheng Cheng and
          Borui Jiang and Chaorui Deng and Yang Zhao and Dong Liu and Yadong Mu and
          Mingkui Tan and Xinggang Wang and Wenyu Liu and Bin Xiao},
  journal={TPAMI},
  year={2019}
}

@inproceedings{zhang2020distribution,
  title={Distribution-aware coordinate representation for human pose estimation},
  author={Zhang, Feng and Zhu, Xiatian and Dai, Hanbin and Ye, Mao and Zhu, Ce},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={7093--7102},
  year={2020}
}

@inproceedings{simon2017hand,
  title={Hand keypoint detection in single images using multiview bootstrapping},
  author={Simon, Tomas and Joo, Hanbyul and Matthews, Iain and Sheikh, Yaser},
  booktitle={Proceedings of the IEEE conference on Computer Vision and Pattern Recognition},
  pages={1145--1153},
  year={2017}
}

Results on CMU Panoptic (MPII+NZSL val set)

Arch	Input Size	PCKh@0.7	AUC	EPE	ckpt	log
pose_hrnetv2_w18_dark	256x256	0.999	0.745	7.77	ckpt	log

Topdown Heatmap + Hrnetv2 + Udp on Panoptic2d¶

@article{WangSCJDZLMTWLX19,
  title={Deep High-Resolution Representation Learning for Visual Recognition},
  author={Jingdong Wang and Ke Sun and Tianheng Cheng and
          Borui Jiang and Chaorui Deng and Yang Zhao and Dong Liu and Yadong Mu and
          Mingkui Tan and Xinggang Wang and Wenyu Liu and Bin Xiao},
  journal={TPAMI},
  year={2019}
}

@InProceedings{Huang_2020_CVPR,
  author = {Huang, Junjie and Zhu, Zheng and Guo, Feng and Huang, Guan},
  title = {The Devil Is in the Details: Delving Into Unbiased Data Processing for Human Pose Estimation},
  booktitle = {The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  month = {June},
  year = {2020}
}

@inproceedings{simon2017hand,
  title={Hand keypoint detection in single images using multiview bootstrapping},
  author={Simon, Tomas and Joo, Hanbyul and Matthews, Iain and Sheikh, Yaser},
  booktitle={Proceedings of the IEEE conference on Computer Vision and Pattern Recognition},
  pages={1145--1153},
  year={2017}
}

Results on CMU Panoptic (MPII+NZSL val set)

Arch	Input Size	PCKh@0.7	AUC	EPE	ckpt	log
pose_hrnetv2_w18_udp	256x256	0.998	0.742	7.84	ckpt	log

Topdown Heatmap + Hrnetv2 on Panoptic2d¶

@article{WangSCJDZLMTWLX19,
  title={Deep High-Resolution Representation Learning for Visual Recognition},
  author={Jingdong Wang and Ke Sun and Tianheng Cheng and
          Borui Jiang and Chaorui Deng and Yang Zhao and Dong Liu and Yadong Mu and
          Mingkui Tan and Xinggang Wang and Wenyu Liu and Bin Xiao},
  journal={TPAMI},
  year={2019}
}

@inproceedings{simon2017hand,
  title={Hand keypoint detection in single images using multiview bootstrapping},
  author={Simon, Tomas and Joo, Hanbyul and Matthews, Iain and Sheikh, Yaser},
  booktitle={Proceedings of the IEEE conference on Computer Vision and Pattern Recognition},
  pages={1145--1153},
  year={2017}
}

Results on CMU Panoptic (MPII+NZSL val set)

Arch	Input Size	PCKh@0.7	AUC	EPE	ckpt	log
pose_hrnetv2_w18	256x256	0.999	0.744	7.79	ckpt	log

AFLW (ICCVW’2011)¶

Topdown Heatmap + Hrnetv2 + Dark on Aflw¶

@article{WangSCJDZLMTWLX19,
  title={Deep High-Resolution Representation Learning for Visual Recognition},
  author={Jingdong Wang and Ke Sun and Tianheng Cheng and
          Borui Jiang and Chaorui Deng and Yang Zhao and Dong Liu and Yadong Mu and
          Mingkui Tan and Xinggang Wang and Wenyu Liu and Bin Xiao},
  journal={TPAMI},
  year={2019}
}

@inproceedings{zhang2020distribution,
  title={Distribution-aware coordinate representation for human pose estimation},
  author={Zhang, Feng and Zhu, Xiatian and Dai, Hanbin and Ye, Mao and Zhu, Ce},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={7093--7102},
  year={2020}
}

AFLW (ICCVW'2011)

@inproceedings{koestinger2011annotated,
  title={Annotated facial landmarks in the wild: A large-scale, real-world database for facial landmark localization},
  author={Koestinger, Martin and Wohlhart, Paul and Roth, Peter M and Bischof, Horst},
  booktitle={2011 IEEE international conference on computer vision workshops (ICCV workshops)},
  pages={2144--2151},
  year={2011},
  organization={IEEE}
}

Results on AFLW dataset

The model is trained on AFLW train and evaluated on AFLW full and frontal.

Arch	Input Size	NME_full	NME_frontal	ckpt	log
pose_hrnetv2_w18_dark	256x256	1.34	1.20	ckpt	log

Topdown Heatmap + Hrnetv2 on Aflw¶

@article{WangSCJDZLMTWLX19,
  title={Deep High-Resolution Representation Learning for Visual Recognition},
  author={Jingdong Wang and Ke Sun and Tianheng Cheng and
          Borui Jiang and Chaorui Deng and Yang Zhao and Dong Liu and Yadong Mu and
          Mingkui Tan and Xinggang Wang and Wenyu Liu and Bin Xiao},
  journal={TPAMI},
  year={2019}
}

AFLW (ICCVW'2011)

@inproceedings{koestinger2011annotated,
  title={Annotated facial landmarks in the wild: A large-scale, real-world database for facial landmark localization},
  author={Koestinger, Martin and Wohlhart, Paul and Roth, Peter M and Bischof, Horst},
  booktitle={2011 IEEE international conference on computer vision workshops (ICCV workshops)},
  pages={2144--2151},
  year={2011},
  organization={IEEE}
}

Results on AFLW dataset

The model is trained on AFLW train and evaluated on AFLW full and frontal.

Arch	Input Size	NME_full	NME_frontal	ckpt	log
pose_hrnetv2_w18	256x256	1.41	1.27	ckpt	log

300W (IMAVIS’2016)¶

Topdown Heatmap + Hrnetv2 on 300w¶

@article{WangSCJDZLMTWLX19,
  title={Deep High-Resolution Representation Learning for Visual Recognition},
  author={Jingdong Wang and Ke Sun and Tianheng Cheng and
          Borui Jiang and Chaorui Deng and Yang Zhao and Dong Liu and Yadong Mu and
          Mingkui Tan and Xinggang Wang and Wenyu Liu and Bin Xiao},
  journal={TPAMI},
  year={2019}
}

300W (IMAVIS'2016)

@article{sagonas2016300,
  title={300 faces in-the-wild challenge: Database and results},
  author={Sagonas, Christos and Antonakos, Epameinondas and Tzimiropoulos, Georgios and Zafeiriou, Stefanos and Pantic, Maja},
  journal={Image and vision computing},
  volume={47},
  pages={3--18},
  year={2016},
  publisher={Elsevier}
}

Results on 300W dataset

The model is trained on 300W train.

Arch	Input Size	NME_common	NME_challenge	NME_full	NME_test	ckpt	log
pose_hrnetv2_w18	256x256	2.86	5.45	3.37	3.97	ckpt	log

FreiHand (ICCV’2019)¶

Topdown Heatmap + Resnet on Freihand2d¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

FreiHand (ICCV'2019)

@inproceedings{zimmermann2019freihand,
  title={Freihand: A dataset for markerless capture of hand pose and shape from single rgb images},
  author={Zimmermann, Christian and Ceylan, Duygu and Yang, Jimei and Russell, Bryan and Argus, Max and Brox, Thomas},
  booktitle={Proceedings of the IEEE International Conference on Computer Vision},
  pages={813--822},
  year={2019}
}

Results on FreiHand val & test set

Set	Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
val	pose_resnet_50	224x224	0.993	0.868	3.25	ckpt	log
test	pose_resnet_50	224x224	0.992	0.868	3.27	ckpt	log

MHP (ACM MM’2018)¶

Associative Embedding + Hrnet on MHP¶

@inproceedings{newell2017associative,
  title={Associative embedding: End-to-end learning for joint detection and grouping},
  author={Newell, Alejandro and Huang, Zhiao and Deng, Jia},
  booktitle={Advances in neural information processing systems},
  pages={2277--2287},
  year={2017}
}

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

MHP (ACM MM'2018)

@inproceedings{zhao2018understanding,
  title={Understanding humans in crowded scenes: Deep nested adversarial learning and a new benchmark for multi-human parsing},
  author={Zhao, Jian and Li, Jianshu and Cheng, Yu and Sim, Terence and Yan, Shuicheng and Feng, Jiashi},
  booktitle={Proceedings of the 26th ACM international conference on Multimedia},
  pages={792--800},
  year={2018}
}

Results on MHP v2.0 validation set without multi-scale test

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
HRNet-w48	512x512	0.583	0.895	0.666	0.656	0.931	ckpt	log

Results on MHP v2.0 validation set with multi-scale test. 3 default scales ([2, 1, 0.5]) are used

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
HRNet-w48	512x512	0.592	0.898	0.673	0.664	0.932	ckpt	log

Topdown Heatmap + Resnet on MHP¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

MHP (ACM MM'2018)

@inproceedings{zhao2018understanding,
  title={Understanding humans in crowded scenes: Deep nested adversarial learning and a new benchmark for multi-human parsing},
  author={Zhao, Jian and Li, Jianshu and Cheng, Yu and Sim, Terence and Yan, Shuicheng and Feng, Jiashi},
  booktitle={Proceedings of the 26th ACM international conference on Multimedia},
  pages={792--800},
  year={2018}
}

Results on MHP v2.0 val set

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_resnet_101	256x192	0.583	0.897	0.669	0.636	0.918	ckpt	log

Note that, the evaluation metric used here is mAP (adapted from COCO), which may be different from the official evaluation codes. Please be cautious if you use the results in papers.

Vinegar Fly (Nature Methods’2019)¶

Topdown Heatmap + Resnet on Fly¶

Vinegar Fly (Nature Methods'2019)

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

@article{pereira2019fast,
  title={Fast animal pose estimation using deep neural networks},
  author={Pereira, Talmo D and Aldarondo, Diego E and Willmore, Lindsay and Kislin, Mikhail and Wang, Samuel S-H and Murthy, Mala and Shaevitz, Joshua W},
  journal={Nature methods},
  volume={16},
  number={1},
  pages={117--125},
  year={2019},
  publisher={Nature Publishing Group}
}

Results on Vinegar Fly test set

Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
pose_resnet_50	192x192	0.996	0.910	2.00	ckpt	log
pose_resnet_101	192x192	0.996	0.912	1.95	ckpt	log
pose_resnet_152	192x192	0.997	0.917	1.78	ckpt	log

CrowdPose (CVPR’2019)¶

Associative Embedding + Higherhrnet on Crowdpose¶

@inproceedings{newell2017associative,
  title={Associative embedding: End-to-end learning for joint detection and grouping},
  author={Newell, Alejandro and Huang, Zhiao and Deng, Jia},
  booktitle={Advances in neural information processing systems},
  pages={2277--2287},
  year={2017}
}

@inproceedings{cheng2020higherhrnet,
  title={HigherHRNet: Scale-Aware Representation Learning for Bottom-Up Human Pose Estimation},
  author={Cheng, Bowen and Xiao, Bin and Wang, Jingdong and Shi, Honghui and Huang, Thomas S and Zhang, Lei},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={5386--5395},
  year={2020}
}

@article{li2018crowdpose,
  title={CrowdPose: Efficient Crowded Scenes Pose Estimation and A New Benchmark},
  author={Li, Jiefeng and Wang, Can and Zhu, Hao and Mao, Yihuan and Fang, Hao-Shu and Lu, Cewu},
  journal={arXiv preprint arXiv:1812.00324},
  year={2018}
}

Results on CrowdPose test without multi-scale test

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AP (E)	AP (M)	AP (H)	ckpt	log
HigherHRNet-w32	512x512	0.655	0.859	0.705	0.728	0.660	0.577	ckpt	log

Results on CrowdPose test with multi-scale test. 2 scales ([2, 1]) are used

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AP (E)	AP (M)	AP (H)	ckpt	log
HigherHRNet-w32	512x512	0.661	0.864	0.710	0.742	0.670	0.566	ckpt	log

Dekr + Hrnet on Crowdpose¶

DEKR (CVPR'2021)

@inproceedings{geng2021bottom,
  title={Bottom-up human pose estimation via disentangled keypoint regression},
  author={Geng, Zigang and Sun, Ke and Xiao, Bin and Zhang, Zhaoxiang and Wang, Jingdong},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={14676--14686},
  year={2021}
}

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

@article{li2018crowdpose,
  title={CrowdPose: Efficient Crowded Scenes Pose Estimation and A New Benchmark},
  author={Li, Jiefeng and Wang, Can and Zhu, Hao and Mao, Yihuan and Fang, Hao-Shu and Lu, Cewu},
  journal={arXiv preprint arXiv:1812.00324},
  year={2018}
}

Results on CrowdPose test without multi-scale test

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
HRNet-w32	512x512	0.663	0.857	0.715	0.719	0.893	ckpt	log
HRNet-w48	640x640	0.682	0.869	0.736	0.742	0.911	ckpt	log

Results on CrowdPose test with multi-scale test. 3 default scales ([2, 1, 0.5]) are used

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt
HRNet-w32*	512x512	0.692	0.874	0.748	0.755	0.926	ckpt
HRNet-w48*	640x640	0.696	0.869	0.749	0.769	0.933	ckpt

* these configs are generally used for evaluation. The training settings are identical to their single-scale counterparts.

Topdown Heatmap + Hrnet on Crowdpose¶

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

@article{li2018crowdpose,
  title={CrowdPose: Efficient Crowded Scenes Pose Estimation and A New Benchmark},
  author={Li, Jiefeng and Wang, Can and Zhu, Hao and Mao, Yihuan and Fang, Hao-Shu and Lu, Cewu},
  journal={arXiv preprint arXiv:1812.00324},
  year={2018}
}

Results on CrowdPose test with YOLOv3 human detector

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AP (E)	AP (M)	AP (H)	ckpt	log
pose_hrnet_w32	256x192	0.675	0.825	0.729	0.770	0.687	0.553	ckpt	log

Topdown Heatmap + Resnet on Crowdpose¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

@article{li2018crowdpose,
  title={CrowdPose: Efficient Crowded Scenes Pose Estimation and A New Benchmark},
  author={Li, Jiefeng and Wang, Can and Zhu, Hao and Mao, Yihuan and Fang, Hao-Shu and Lu, Cewu},
  journal={arXiv preprint arXiv:1812.00324},
  year={2018}
}

Results on CrowdPose test with YOLOv3 human detector

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AP (E)	AP (M)	AP (H)	ckpt	log
pose_resnet_50	256x192	0.637	0.808	0.692	0.739	0.650	0.506	ckpt	log
pose_resnet_101	256x192	0.647	0.810	0.703	0.744	0.658	0.522	ckpt	log
pose_resnet_101	320x256	0.661	0.821	0.714	0.759	0.671	0.536	ckpt	log
pose_resnet_152	256x192	0.656	0.818	0.712	0.754	0.666	0.532	ckpt	log

Halpe (CVPR’2020)¶

Topdown Heatmap + Hrnet + Dark on Halpe¶

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

@inproceedings{zhang2020distribution,
  title={Distribution-aware coordinate representation for human pose estimation},
  author={Zhang, Feng and Zhu, Xiatian and Dai, Hanbin and Ye, Mao and Zhu, Ce},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={7093--7102},
  year={2020}
}

Halpe (CVPR'2020)

@inproceedings{li2020pastanet,
  title={PaStaNet: Toward Human Activity Knowledge Engine},
  author={Li, Yong-Lu and Xu, Liang and Liu, Xinpeng and Huang, Xijie and Xu, Yue and Wang, Shiyi and Fang, Hao-Shu and Ma, Ze and Chen, Mingyang and Lu, Cewu},
  booktitle={CVPR},
  year={2020}
}

Results on Halpe v1.0 val with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	Whole AP	Whole AR	ckpt	log
pose_hrnet_w48_dark+	384x288	0.527	0.620	ckpt	log

Note: + means the model is first pre-trained on original COCO dataset, and then fine-tuned on Halpe dataset. We find this will lead to better performance.

OCHuman (CVPR’2019)¶

Topdown Heatmap + Resnet on Ochuman¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

OCHuman (CVPR'2019)

@inproceedings{zhang2019pose2seg,
  title={Pose2seg: Detection free human instance segmentation},
  author={Zhang, Song-Hai and Li, Ruilong and Dong, Xin and Rosin, Paul and Cai, Zixi and Han, Xi and Yang, Dingcheng and Huang, Haozhi and Hu, Shi-Min},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={889--898},
  year={2019}
}

Results on OCHuman test dataset with ground-truth bounding boxes

Following the common setting, the models are trained on COCO train dataset, and evaluate on OCHuman dataset.

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_resnet_50	256x192	0.546	0.726	0.593	0.592	0.755	ckpt	log
pose_resnet_50	384x288	0.539	0.723	0.574	0.588	0.756	ckpt	log
pose_resnet_101	256x192	0.559	0.724	0.606	0.605	0.751	ckpt	log
pose_resnet_101	384x288	0.571	0.715	0.615	0.615	0.748	ckpt	log
pose_resnet_152	256x192	0.570	0.725	0.617	0.616	0.754	ckpt	log
pose_resnet_152	384x288	0.582	0.723	0.627	0.627	0.752	ckpt	log

Topdown Heatmap + Hrnet on Ochuman¶

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

OCHuman (CVPR'2019)

@inproceedings{zhang2019pose2seg,
  title={Pose2seg: Detection free human instance segmentation},
  author={Zhang, Song-Hai and Li, Ruilong and Dong, Xin and Rosin, Paul and Cai, Zixi and Han, Xi and Yang, Dingcheng and Huang, Haozhi and Hu, Shi-Min},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={889--898},
  year={2019}
}

Results on OCHuman test dataset with ground-truth bounding boxes

Following the common setting, the models are trained on COCO train dataset, and evaluate on OCHuman dataset.

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_hrnet_w32	256x192	0.591	0.748	0.641	0.631	0.775	ckpt	log
pose_hrnet_w32	384x288	0.606	0.748	0.650	0.647	0.776	ckpt	log
pose_hrnet_w48	256x192	0.611	0.752	0.663	0.648	0.778	ckpt	log
pose_hrnet_w48	384x288	0.616	0.749	0.663	0.653	0.773	ckpt	log

AP-10K (NeurIPS’2021)¶

Topdown Heatmap + Hrnet on Ap10k¶

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

AP-10K (NeurIPS'2021)

@misc{yu2021ap10k,
      title={AP-10K: A Benchmark for Animal Pose Estimation in the Wild},
      author={Hang Yu and Yufei Xu and Jing Zhang and Wei Zhao and Ziyu Guan and Dacheng Tao},
      year={2021},
      eprint={2108.12617},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}

Results on AP-10K validation set

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AP^M	AP^L	ckpt	log
pose_hrnet_w32	256x256	0.722	0.939	0.787	0.555	0.730	ckpt	log
pose_hrnet_w48	256x256	0.731	0.937	0.804	0.574	0.738	ckpt	log

Topdown Heatmap + Resnet on Ap10k¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

AP-10K (NeurIPS'2021)

@misc{yu2021ap10k,
      title={AP-10K: A Benchmark for Animal Pose Estimation in the Wild},
      author={Hang Yu and Yufei Xu and Jing Zhang and Wei Zhao and Ziyu Guan and Dacheng Tao},
      year={2021},
      eprint={2108.12617},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}

Results on AP-10K validation set

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AP^M	AP^L	ckpt	log
pose_resnet_50	256x256	0.681	0.923	0.740	0.510	0.688	ckpt	log
pose_resnet_101	256x256	0.681	0.922	0.742	0.534	0.688	ckpt	log

Desert Locust (Elife’2019)¶

Topdown Heatmap + Resnet on Locust¶

Desert Locust (Elife'2019)

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

@article{graving2019deepposekit,
  title={DeepPoseKit, a software toolkit for fast and robust animal pose estimation using deep learning},
  author={Graving, Jacob M and Chae, Daniel and Naik, Hemal and Li, Liang and Koger, Benjamin and Costelloe, Blair R and Couzin, Iain D},
  journal={Elife},
  volume={8},
  pages={e47994},
  year={2019},
  publisher={eLife Sciences Publications Limited}
}

Results on Desert Locust test set

Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
pose_resnet_50	160x160	0.999	0.899	2.27	ckpt	log
pose_resnet_101	160x160	0.999	0.907	2.03	ckpt	log
pose_resnet_152	160x160	1.000	0.926	1.48	ckpt	log

COCO-WholeBody-Face (ECCV’2020)¶

Topdown Heatmap + Mobilenetv2 + Coco + Wholebody on Coco_wholebody_face¶

@inproceedings{sandler2018mobilenetv2,
  title={Mobilenetv2: Inverted residuals and linear bottlenecks},
  author={Sandler, Mark and Howard, Andrew and Zhu, Menglong and Zhmoginov, Andrey and Chen, Liang-Chieh},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={4510--4520},
  year={2018}
}

@inproceedings{jin2020whole,
  title={Whole-Body Human Pose Estimation in the Wild},
  author={Jin, Sheng and Xu, Lumin and Xu, Jin and Wang, Can and Liu, Wentao and Qian, Chen and Ouyang, Wanli and Luo, Ping},
  booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
  year={2020}
}

Results on COCO-WholeBody-Face val set

Arch	Input Size	NME	ckpt	log
pose_mobilenetv2	256x256	0.0612	ckpt	log

Topdown Heatmap + Resnet + Coco + Wholebody on Coco_wholebody_face¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

@inproceedings{jin2020whole,
  title={Whole-Body Human Pose Estimation in the Wild},
  author={Jin, Sheng and Xu, Lumin and Xu, Jin and Wang, Can and Liu, Wentao and Qian, Chen and Ouyang, Wanli and Luo, Ping},
  booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
  year={2020}
}

Results on COCO-WholeBody-Face val set

Arch	Input Size	NME	ckpt	log
pose_res50	256x256	0.0566	ckpt	log

Topdown Heatmap + Hrnetv2 + Coco + Wholebody on Coco_wholebody_face¶

@article{WangSCJDZLMTWLX19,
  title={Deep High-Resolution Representation Learning for Visual Recognition},
  author={Jingdong Wang and Ke Sun and Tianheng Cheng and
          Borui Jiang and Chaorui Deng and Yang Zhao and Dong Liu and Yadong Mu and
          Mingkui Tan and Xinggang Wang and Wenyu Liu and Bin Xiao},
  journal={TPAMI},
  year={2019}
}

@inproceedings{jin2020whole,
  title={Whole-Body Human Pose Estimation in the Wild},
  author={Jin, Sheng and Xu, Lumin and Xu, Jin and Wang, Can and Liu, Wentao and Qian, Chen and Ouyang, Wanli and Luo, Ping},
  booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
  year={2020}
}

Results on COCO-WholeBody-Face val set

Arch	Input Size	NME	ckpt	log
pose_hrnetv2_w18	256x256	0.0569	ckpt	log

Topdown Heatmap + Scnet + Coco + Wholebody on Coco_wholebody_face¶

@inproceedings{liu2020improving,
  title={Improving Convolutional Networks with Self-Calibrated Convolutions},
  author={Liu, Jiang-Jiang and Hou, Qibin and Cheng, Ming-Ming and Wang, Changhu and Feng, Jiashi},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={10096--10105},
  year={2020}
}

@inproceedings{jin2020whole,
  title={Whole-Body Human Pose Estimation in the Wild},
  author={Jin, Sheng and Xu, Lumin and Xu, Jin and Wang, Can and Liu, Wentao and Qian, Chen and Ouyang, Wanli and Luo, Ping},
  booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
  year={2020}
}

Results on COCO-WholeBody-Face val set

Arch	Input Size	NME	ckpt	log
pose_scnet_50	256x256	0.0565	ckpt	log

Topdown Heatmap + Hourglass + Coco + Wholebody on Coco_wholebody_face¶

@inproceedings{newell2016stacked,
  title={Stacked hourglass networks for human pose estimation},
  author={Newell, Alejandro and Yang, Kaiyu and Deng, Jia},
  booktitle={European conference on computer vision},
  pages={483--499},
  year={2016},
  organization={Springer}
}

@inproceedings{jin2020whole,
  title={Whole-Body Human Pose Estimation in the Wild},
  author={Jin, Sheng and Xu, Lumin and Xu, Jin and Wang, Can and Liu, Wentao and Qian, Chen and Ouyang, Wanli and Luo, Ping},
  booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
  year={2020}
}

Results on COCO-WholeBody-Face val set

Arch	Input Size	NME	ckpt	log
pose_hourglass_52	256x256	0.0586	ckpt	log

Topdown Heatmap + Hrnetv2 + Dark + Coco + Wholebody on Coco_wholebody_face¶

@article{WangSCJDZLMTWLX19,
  title={Deep High-Resolution Representation Learning for Visual Recognition},
  author={Jingdong Wang and Ke Sun and Tianheng Cheng and
          Borui Jiang and Chaorui Deng and Yang Zhao and Dong Liu and Yadong Mu and
          Mingkui Tan and Xinggang Wang and Wenyu Liu and Bin Xiao},
  journal={TPAMI},
  year={2019}
}

@inproceedings{zhang2020distribution,
  title={Distribution-aware coordinate representation for human pose estimation},
  author={Zhang, Feng and Zhu, Xiatian and Dai, Hanbin and Ye, Mao and Zhu, Ce},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={7093--7102},
  year={2020}
}

SimpleBaseline3D (ICCV'2017)

@inproceedings{jin2020whole,
  title={Whole-Body Human Pose Estimation in the Wild},
  author={Jin, Sheng and Xu, Lumin and Xu, Jin and Wang, Can and Liu, Wentao and Qian, Chen and Ouyang, Wanli and Luo, Ping},
  booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
  year={2020}
}

Results on COCO-WholeBody-Face val set

Arch	Input Size	NME	ckpt	log
pose_hrnetv2_w18_dark	256x256	0.0513	ckpt	log

MPI-INF-3DHP (3DV’2017)¶

Pose Lift + Simplebaseline3d on Mpi_inf_3dhp¶

@inproceedings{martinez_2017_3dbaseline,
  title={A simple yet effective baseline for 3d human pose estimation},
  author={Martinez, Julieta and Hossain, Rayat and Romero, Javier and Little, James J.},
  booktitle={ICCV},
  year={2017}
}

MPI-INF-3DHP (3DV'2017)

@inproceedings{mono-3dhp2017,
  author = {Mehta, Dushyant and Rhodin, Helge and Casas, Dan and Fua, Pascal and Sotnychenko, Oleksandr and Xu, Weipeng and Theobalt, Christian},
  title = {Monocular 3D Human Pose Estimation In The Wild Using Improved CNN Supervision},
  booktitle = {3D Vision (3DV), 2017 Fifth International Conference on},
  url = {http://gvv.mpi-inf.mpg.de/3dhp_dataset},
  year = {2017},
  organization={IEEE},
  doi={10.1109/3dv.2017.00064},
}

Results on MPI-INF-3DHP dataset with ground truth 2D detections

Arch	MPJPE	P-MPJPE	3DPCK	3DAUC	ckpt	log
simple_baseline_3d_tcn¹	84.3	53.2	85.0	52.0	ckpt	log

¹ Differing from the original paper, we didn’t apply the max-norm constraint because we found this led to a better convergence and performance.

Video Pose Lift + Videopose3d on Mpi_inf_3dhp¶

VideoPose3D (CVPR'2019)

@inproceedings{pavllo20193d,
  title={3d human pose estimation in video with temporal convolutions and semi-supervised training},
  author={Pavllo, Dario and Feichtenhofer, Christoph and Grangier, David and Auli, Michael},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={7753--7762},
  year={2019}
}

MPI-INF-3DHP (3DV'2017)

@inproceedings{mono-3dhp2017,
  author = {Mehta, Dushyant and Rhodin, Helge and Casas, Dan and Fua, Pascal and Sotnychenko, Oleksandr and Xu, Weipeng and Theobalt, Christian},
  title = {Monocular 3D Human Pose Estimation In The Wild Using Improved CNN Supervision},
  booktitle = {3D Vision (3DV), 2017 Fifth International Conference on},
  url = {http://gvv.mpi-inf.mpg.de/3dhp_dataset},
  year = {2017},
  organization={IEEE},
  doi={10.1109/3dv.2017.00064},
}

Results on MPI-INF-3DHP dataset with ground truth 2D detections, supervised training

Arch	Receptive Field	MPJPE	P-MPJPE	3DPCK	3DAUC	ckpt	log
VideoPose3D	1	58.3	40.6	94.1	63.1	ckpt	log

AI Challenger (ArXiv’2017)¶

Associative Embedding + Hrnet on Aic¶

@inproceedings{newell2017associative,
  title={Associative embedding: End-to-end learning for joint detection and grouping},
  author={Newell, Alejandro and Huang, Zhiao and Deng, Jia},
  booktitle={Advances in neural information processing systems},
  pages={2277--2287},
  year={2017}
}

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

@article{wu2017ai,
  title={Ai challenger: A large-scale dataset for going deeper in image understanding},
  author={Wu, Jiahong and Zheng, He and Zhao, Bo and Li, Yixin and Yan, Baoming and Liang, Rui and Wang, Wenjia and Zhou, Shipei and Lin, Guosen and Fu, Yanwei and others},
  journal={arXiv preprint arXiv:1711.06475},
  year={2017}
}

Results on AIC validation set without multi-scale test

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
HRNet-w32	512x512	0.303	0.697	0.225	0.373	0.755	ckpt	log

Results on AIC validation set with multi-scale test. 3 default scales ([2, 1, 0.5]) are used

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
HRNet-w32	512x512	0.318	0.717	0.246	0.379	0.764	ckpt	log

Associative Embedding + Higherhrnet on Aic¶

@inproceedings{newell2017associative,
  title={Associative embedding: End-to-end learning for joint detection and grouping},
  author={Newell, Alejandro and Huang, Zhiao and Deng, Jia},
  booktitle={Advances in neural information processing systems},
  pages={2277--2287},
  year={2017}
}

@inproceedings{cheng2020higherhrnet,
  title={HigherHRNet: Scale-Aware Representation Learning for Bottom-Up Human Pose Estimation},
  author={Cheng, Bowen and Xiao, Bin and Wang, Jingdong and Shi, Honghui and Huang, Thomas S and Zhang, Lei},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={5386--5395},
  year={2020}
}

@article{wu2017ai,
  title={Ai challenger: A large-scale dataset for going deeper in image understanding},
  author={Wu, Jiahong and Zheng, He and Zhao, Bo and Li, Yixin and Yan, Baoming and Liang, Rui and Wang, Wenjia and Zhou, Shipei and Lin, Guosen and Fu, Yanwei and others},
  journal={arXiv preprint arXiv:1711.06475},
  year={2017}
}

Results on AIC validation set without multi-scale test

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
HigherHRNet-w32	512x512	0.315	0.710	0.243	0.379	0.757	ckpt	log

Results on AIC validation set with multi-scale test. 3 default scales ([2, 1, 0.5]) are used

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
HigherHRNet-w32	512x512	0.323	0.718	0.254	0.379	0.758	ckpt	log

Topdown Heatmap + Hrnet on Aic¶

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

@article{wu2017ai,
  title={Ai challenger: A large-scale dataset for going deeper in image understanding},
  author={Wu, Jiahong and Zheng, He and Zhao, Bo and Li, Yixin and Yan, Baoming and Liang, Rui and Wang, Wenjia and Zhou, Shipei and Lin, Guosen and Fu, Yanwei and others},
  journal={arXiv preprint arXiv:1711.06475},
  year={2017}
}

Results on AIC val set with ground-truth bounding boxes

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_hrnet_w32	256x192	0.323	0.762	0.219	0.366	0.789	ckpt	log

Topdown Heatmap + Resnet on Aic¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

@article{wu2017ai,
  title={Ai challenger: A large-scale dataset for going deeper in image understanding},
  author={Wu, Jiahong and Zheng, He and Zhao, Bo and Li, Yixin and Yan, Baoming and Liang, Rui and Wang, Wenjia and Zhou, Shipei and Lin, Guosen and Fu, Yanwei and others},
  journal={arXiv preprint arXiv:1711.06475},
  year={2017}
}

Results on AIC val set with ground-truth bounding boxes

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_resnet_101	256x192	0.294	0.736	0.174	0.337	0.763	ckpt	log

MPII-TRB (ICCV’2019)¶

Topdown Heatmap + Resnet + Mpii on Mpii_trb¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

MPII-TRB (ICCV'2019)

@inproceedings{duan2019trb,
  title={TRB: A Novel Triplet Representation for Understanding 2D Human Body},
  author={Duan, Haodong and Lin, Kwan-Yee and Jin, Sheng and Liu, Wentao and Qian, Chen and Ouyang, Wanli},
  booktitle={Proceedings of the IEEE International Conference on Computer Vision},
  pages={9479--9488},
  year={2019}
}

Results on MPII-TRB val set

Arch	Input Size	Skeleton Acc	Contour Acc	Mean Acc	ckpt	log
pose_resnet_50	256x256	0.887	0.858	0.868	ckpt	log
pose_resnet_101	256x256	0.890	0.863	0.873	ckpt	log
pose_resnet_152	256x256	0.897	0.868	0.879	ckpt	log

WFLW (CVPR’2018)¶

Deeppose + Resnet on WFLW¶

@inproceedings{toshev2014deeppose,
  title={Deeppose: Human pose estimation via deep neural networks},
  author={Toshev, Alexander and Szegedy, Christian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={1653--1660},
  year={2014}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

@inproceedings{wu2018look,
  title={Look at boundary: A boundary-aware face alignment algorithm},
  author={Wu, Wayne and Qian, Chen and Yang, Shuo and Wang, Quan and Cai, Yici and Zhou, Qiang},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={2129--2138},
  year={2018}
}

Results on WFLW dataset

The model is trained on WFLW train.

Arch	Input Size	NME_test	NME_pose	NME_illumination	NME_occlusion	NME_blur	NME_makeup	NME_expression	ckpt	log
deeppose_res50	256x256	4.85	8.50	4.81	5.69	5.45	4.82	5.20	ckpt	log

Deeppose + Resnet + Softwingloss on WFLW¶

@inproceedings{toshev2014deeppose,
  title={Deeppose: Human pose estimation via deep neural networks},
  author={Toshev, Alexander and Szegedy, Christian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={1653--1660},
  year={2014}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

SoftWingloss (TIP'2021)

@article{lin2021structure,
  title={Structure-Coherent Deep Feature Learning for Robust Face Alignment},
  author={Lin, Chunze and Zhu, Beier and Wang, Quan and Liao, Renjie and Qian, Chen and Lu, Jiwen and Zhou, Jie},
  journal={IEEE Transactions on Image Processing},
  year={2021},
  publisher={IEEE}
}

@inproceedings{wu2018look,
  title={Look at boundary: A boundary-aware face alignment algorithm},
  author={Wu, Wayne and Qian, Chen and Yang, Shuo and Wang, Quan and Cai, Yici and Zhou, Qiang},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={2129--2138},
  year={2018}
}

Results on WFLW dataset

The model is trained on WFLW train.

Arch	Input Size	NME_test	NME_pose	NME_illumination	NME_occlusion	NME_blur	NME_makeup	NME_expression	ckpt	log
deeppose_res50_softwingloss	256x256	4.41	7.77	4.37	5.27	5.01	4.36	4.70	ckpt	log

Deeppose + Resnet + Wingloss on WFLW¶

@inproceedings{toshev2014deeppose,
  title={Deeppose: Human pose estimation via deep neural networks},
  author={Toshev, Alexander and Szegedy, Christian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={1653--1660},
  year={2014}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

Wingloss (CVPR'2018)

@inproceedings{feng2018wing,
  title={Wing Loss for Robust Facial Landmark Localisation with Convolutional Neural Networks},
  author={Feng, Zhen-Hua and Kittler, Josef and Awais, Muhammad and Huber, Patrik and Wu, Xiao-Jun},
  booktitle={Computer Vision and Pattern Recognition (CVPR), 2018 IEEE Conference on},
  year={2018},
  pages ={2235-2245},
  organization={IEEE}
}

@inproceedings{wu2018look,
  title={Look at boundary: A boundary-aware face alignment algorithm},
  author={Wu, Wayne and Qian, Chen and Yang, Shuo and Wang, Quan and Cai, Yici and Zhou, Qiang},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={2129--2138},
  year={2018}
}

Results on WFLW dataset

The model is trained on WFLW train.

Arch	Input Size	NME_test	NME_pose	NME_illumination	NME_occlusion	NME_blur	NME_makeup	NME_expression	ckpt	log
deeppose_res50_wingloss	256x256	4.64	8.25	4.59	5.56	5.26	4.59	5.07	ckpt	log

Topdown Heatmap + Hrnetv2 + Awing on WFLW¶

AdaptiveWingloss (ICCV'2019)

@article{WangSCJDZLMTWLX19,
  title={Deep High-Resolution Representation Learning for Visual Recognition},
  author={Jingdong Wang and Ke Sun and Tianheng Cheng and
          Borui Jiang and Chaorui Deng and Yang Zhao and Dong Liu and Yadong Mu and
          Mingkui Tan and Xinggang Wang and Wenyu Liu and Bin Xiao},
  journal={TPAMI},
  year={2019}
}

@inproceedings{wang2019adaptive,
  title={Adaptive wing loss for robust face alignment via heatmap regression},
  author={Wang, Xinyao and Bo, Liefeng and Fuxin, Li},
  booktitle={Proceedings of the IEEE/CVF international conference on computer vision},
  pages={6971--6981},
  year={2019}
}

@inproceedings{wu2018look,
  title={Look at boundary: A boundary-aware face alignment algorithm},
  author={Wu, Wayne and Qian, Chen and Yang, Shuo and Wang, Quan and Cai, Yici and Zhou, Qiang},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={2129--2138},
  year={2018}
}

Results on WFLW dataset

The model is trained on WFLW train.

Arch	Input Size	NME_test	NME_pose	NME_illumination	NME_occlusion	NME_blur	NME_makeup	NME_expression	ckpt	log
pose_hrnetv2_w18_awing	256x256	4.02	6.94	3.96	4.78	4.59	3.85	4.28	ckpt	log

Topdown Heatmap + Hrnetv2 + Dark on WFLW¶

@article{WangSCJDZLMTWLX19,
  title={Deep High-Resolution Representation Learning for Visual Recognition},
  author={Jingdong Wang and Ke Sun and Tianheng Cheng and
          Borui Jiang and Chaorui Deng and Yang Zhao and Dong Liu and Yadong Mu and
          Mingkui Tan and Xinggang Wang and Wenyu Liu and Bin Xiao},
  journal={TPAMI},
  year={2019}
}

@inproceedings{zhang2020distribution,
  title={Distribution-aware coordinate representation for human pose estimation},
  author={Zhang, Feng and Zhu, Xiatian and Dai, Hanbin and Ye, Mao and Zhu, Ce},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={7093--7102},
  year={2020}
}

@inproceedings{wu2018look,
  title={Look at boundary: A boundary-aware face alignment algorithm},
  author={Wu, Wayne and Qian, Chen and Yang, Shuo and Wang, Quan and Cai, Yici and Zhou, Qiang},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={2129--2138},
  year={2018}
}

Results on WFLW dataset

The model is trained on WFLW train.

Arch	Input Size	NME_test	NME_pose	NME_illumination	NME_occlusion	NME_blur	NME_makeup	NME_expression	ckpt	log
pose_hrnetv2_w18_dark	256x256	3.98	6.99	3.96	4.78	4.57	3.87	4.30	ckpt	log

Topdown Heatmap + Hrnetv2 on WFLW¶

@article{WangSCJDZLMTWLX19,
  title={Deep High-Resolution Representation Learning for Visual Recognition},
  author={Jingdong Wang and Ke Sun and Tianheng Cheng and
          Borui Jiang and Chaorui Deng and Yang Zhao and Dong Liu and Yadong Mu and
          Mingkui Tan and Xinggang Wang and Wenyu Liu and Bin Xiao},
  journal={TPAMI},
  year={2019}
}

@inproceedings{wu2018look,
  title={Look at boundary: A boundary-aware face alignment algorithm},
  author={Wu, Wayne and Qian, Chen and Yang, Shuo and Wang, Quan and Cai, Yici and Zhou, Qiang},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={2129--2138},
  year={2018}
}

Results on WFLW dataset

The model is trained on WFLW train.

Arch	Input Size	NME_test	NME_pose	NME_illumination	NME_occlusion	NME_blur	NME_makeup	NME_expression	ckpt	log
pose_hrnetv2_w18	256x256	4.06	6.98	3.99	4.83	4.59	3.92	4.33	ckpt	log

DeepFashion2 (CVPR’2019)¶

Topdown Heatmap + Resnet on Deepfashion2¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

DeepFashion2 (CVPR'2019)

@article{DeepFashion2,
  author = {Yuying Ge and Ruimao Zhang and Lingyun Wu and Xiaogang Wang and Xiaoou Tang and Ping Luo},
  title={A Versatile Benchmark for Detection, Pose Estimation, Segmentation and Re-Identification of Clothing Images},
  journal={CVPR},
  year={2019}
}

Results on DeepFashion2 val set

Set	Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
short_sleeved_shirt	pose_resnet_50	256x256	0.988	0.703	10.2	ckpt	log
long_sleeved_shirt	pose_resnet_50	256x256	0.973	0.587	16.5	ckpt	log
short_sleeved_outwear	pose_resnet_50	256x256	0.966	0.408	24.0	ckpt	log
long_sleeved_outwear	pose_resnet_50	256x256	0.987	0.517	18.1	ckpt	log
vest	pose_resnet_50	256x256	0.981	0.643	12.7	ckpt	log
sling	pose_resnet_50	256x256	0.940	0.557	21.6	ckpt	log
shorts	pose_resnet_50	256x256	0.975	0.682	12.4	ckpt	log
trousers	pose_resnet_50	256x256	0.973	0.625	14.8	ckpt	log
skirt	pose_resnet_50	256x256	0.952	0.653	16.6	ckpt	log
short_sleeved_dress	pose_resnet_50	256x256	0.980	0.603	15.6	ckpt	log
long_sleeved_dress	pose_resnet_50	256x256	0.976	0.518	20.1	ckpt	log
vest_dress	pose_resnet_50	256x256	0.980	0.600	16.0	ckpt	log
sling_dress	pose_resnet_50	256x256	0.967	0.544	19.5	ckpt	log

PoseTrack18 (CVPR’2018)¶

Topdown Heatmap + Hrnet on Posetrack18¶

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

@inproceedings{andriluka2018posetrack,
  title={Posetrack: A benchmark for human pose estimation and tracking},
  author={Andriluka, Mykhaylo and Iqbal, Umar and Insafutdinov, Eldar and Pishchulin, Leonid and Milan, Anton and Gall, Juergen and Schiele, Bernt},
  booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
  pages={5167--5176},
  year={2018}
}

Results on PoseTrack2018 val with ground-truth bounding boxes

Arch	Input Size	Head	Shou	Elb	Wri	Hip	Knee	Ankl	Total	ckpt	log
pose_hrnet_w32	256x192	87.4	88.6	84.3	78.5	79.7	81.8	78.8	83.0	ckpt	log
pose_hrnet_w32	384x288	87.0	88.8	85.0	80.1	80.5	82.6	79.4	83.6	ckpt	log
pose_hrnet_w48	256x192	88.2	90.1	85.8	80.8	80.7	83.3	80.3	84.4	ckpt	log
pose_hrnet_w48	384x288	87.8	90.0	85.9	81.3	81.1	83.3	80.9	84.5	ckpt	log

The models are first pre-trained on COCO dataset, and then fine-tuned on PoseTrack18.

Results on PoseTrack2018 val with MMDetection pre-trained Cascade R-CNN (X-101-64x4d-FPN) human detector

Arch	Input Size	Head	Shou	Elb	Wri	Hip	Knee	Ankl	Total	ckpt	log
pose_hrnet_w32	256x192	78.0	82.9	79.5	73.8	76.9	76.6	70.2	76.9	ckpt	log
pose_hrnet_w32	384x288	79.9	83.6	80.4	74.5	74.8	76.1	70.5	77.3	ckpt	log
pose_hrnet_w48	256x192	80.1	83.4	80.6	74.8	74.3	76.8	70.4	77.4	ckpt	log
pose_hrnet_w48	384x288	80.2	83.8	80.9	75.2	74.7	76.7	71.7	77.8	ckpt	log

The models are first pre-trained on COCO dataset, and then fine-tuned on PoseTrack18.

Topdown Heatmap + Resnet on Posetrack18¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

PoseWarper (NeurIPS'2019)

@inproceedings{andriluka2018posetrack,
  title={Posetrack: A benchmark for human pose estimation and tracking},
  author={Andriluka, Mykhaylo and Iqbal, Umar and Insafutdinov, Eldar and Pishchulin, Leonid and Milan, Anton and Gall, Juergen and Schiele, Bernt},
  booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
  pages={5167--5176},
  year={2018}
}

Results on PoseTrack2018 val with ground-truth bounding boxes

Arch	Input Size	Head	Shou	Elb	Wri	Hip	Knee	Ankl	Total	ckpt	log
pose_resnet_50	256x192	86.5	87.5	82.3	75.6	79.9	78.6	74.0	81.0	ckpt	log

The models are first pre-trained on COCO dataset, and then fine-tuned on PoseTrack18.

Results on PoseTrack2018 val with MMDetection pre-trained Cascade R-CNN (X-101-64x4d-FPN) human detector

Arch	Input Size	Head	Shou	Elb	Wri	Hip	Knee	Ankl	Total	ckpt	log
pose_resnet_50	256x192	78.9	81.9	77.8	70.8	75.3	73.2	66.4	75.2	ckpt	log

The models are first pre-trained on COCO dataset, and then fine-tuned on PoseTrack18.

Posewarper + Hrnet + Posetrack18 on Posetrack18¶

@inproceedings{NIPS2019_gberta,
title = {Learning Temporal Pose Estimation from Sparsely Labeled Videos},
author = {Bertasius, Gedas and Feichtenhofer, Christoph, and Tran, Du and Shi, Jianbo, and Torresani, Lorenzo},
booktitle = {Advances in Neural Information Processing Systems 33},
year = {2019},
}

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

@inproceedings{andriluka2018posetrack,
  title={Posetrack: A benchmark for human pose estimation and tracking},
  author={Andriluka, Mykhaylo and Iqbal, Umar and Insafutdinov, Eldar and Pishchulin, Leonid and Milan, Anton and Gall, Juergen and Schiele, Bernt},
  booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
  pages={5167--5176},
  year={2018}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Note that the training of PoseWarper can be split into two stages.

The first-stage is trained with the pre-trained checkpoint on COCO dataset, and the main backbone is fine-tuned on PoseTrack18 in a single-frame setting.

The second-stage is trained with the last checkpoint from the first stage, and the warping offsets are learned in a multi-frame setting while the backbone is frozen.

Results on PoseTrack2018 val with ground-truth bounding boxes

Arch	Input Size	Head	Shou	Elb	Wri	Hip	Knee	Ankl	Total	ckpt	log
pose_hrnet_w48	384x288	88.2	90.3	86.1	81.6	81.8	83.8	81.5	85.0	ckpt	log

Results on PoseTrack2018 val with precomputed human bounding boxes from PoseWarper supplementary data files from this link¹.

Arch	Input Size	Head	Shou	Elb	Wri	Hip	Knee	Ankl	Total	ckpt	log
pose_hrnet_w48	384x288	81.8	85.6	82.7	77.2	76.8	79.0	74.4	79.8	ckpt	log

¹ Please download the precomputed human bounding boxes on PoseTrack2018 val from $PoseWarper_supp_files/posetrack18_precomputed_boxes/val_boxes.json and place it here: $mmpose/data/posetrack18/posetrack18_precomputed_boxes/val_boxes.json to be consistent with the config. Please refer to DATA Preparation for more detail about data preparation.

COCO-WholeBody-Hand (ECCV’2020)¶

Topdown Heatmap + Hrnetv2 + Dark + Coco + Wholebody on Coco_wholebody_hand¶

@article{WangSCJDZLMTWLX19,
  title={Deep High-Resolution Representation Learning for Visual Recognition},
  author={Jingdong Wang and Ke Sun and Tianheng Cheng and
          Borui Jiang and Chaorui Deng and Yang Zhao and Dong Liu and Yadong Mu and
          Mingkui Tan and Xinggang Wang and Wenyu Liu and Bin Xiao},
  journal={TPAMI},
  year={2019}
}

@inproceedings{zhang2020distribution,
  title={Distribution-aware coordinate representation for human pose estimation},
  author={Zhang, Feng and Zhu, Xiatian and Dai, Hanbin and Ye, Mao and Zhu, Ce},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={7093--7102},
  year={2020}
}

@inproceedings{jin2020whole,
  title={Whole-Body Human Pose Estimation in the Wild},
  author={Jin, Sheng and Xu, Lumin and Xu, Jin and Wang, Can and Liu, Wentao and Qian, Chen and Ouyang, Wanli and Luo, Ping},
  booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
  year={2020}
}

Results on COCO-WholeBody-Hand val set

Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
pose_hrnetv2_w18_dark	256x256	0.814	0.840	4.37	ckpt	log

Topdown Heatmap + Mobilenetv2 + Coco + Wholebody on Coco_wholebody_hand¶

@inproceedings{sandler2018mobilenetv2,
  title={Mobilenetv2: Inverted residuals and linear bottlenecks},
  author={Sandler, Mark and Howard, Andrew and Zhu, Menglong and Zhmoginov, Andrey and Chen, Liang-Chieh},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={4510--4520},
  year={2018}
}

@inproceedings{jin2020whole,
  title={Whole-Body Human Pose Estimation in the Wild},
  author={Jin, Sheng and Xu, Lumin and Xu, Jin and Wang, Can and Liu, Wentao and Qian, Chen and Ouyang, Wanli and Luo, Ping},
  booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
  year={2020}
}

Results on COCO-WholeBody-Hand val set

Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
pose_mobilenetv2	256x256	0.795	0.829	4.77	ckpt	log

Topdown Heatmap + Litehrnet + Coco + Wholebody on Coco_wholebody_hand¶

LiteHRNet (CVPR'2021)

@inproceedings{Yulitehrnet21,
  title={Lite-HRNet: A Lightweight High-Resolution Network},
  author={Yu, Changqian and Xiao, Bin and Gao, Changxin and Yuan, Lu and Zhang, Lei and Sang, Nong and Wang, Jingdong},
  booktitle={CVPR},
  year={2021}
}

@inproceedings{jin2020whole,
  title={Whole-Body Human Pose Estimation in the Wild},
  author={Jin, Sheng and Xu, Lumin and Xu, Jin and Wang, Can and Liu, Wentao and Qian, Chen and Ouyang, Wanli and Luo, Ping},
  booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
  year={2020}
}

Results on COCO-WholeBody-Hand val set

Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
LiteHRNet-18	256x256	0.795	0.830	4.77	ckpt	log

Topdown Heatmap + Hourglass + Coco + Wholebody on Coco_wholebody_hand¶

@inproceedings{newell2016stacked,
  title={Stacked hourglass networks for human pose estimation},
  author={Newell, Alejandro and Yang, Kaiyu and Deng, Jia},
  booktitle={European conference on computer vision},
  pages={483--499},
  year={2016},
  organization={Springer}
}

@inproceedings{jin2020whole,
  title={Whole-Body Human Pose Estimation in the Wild},
  author={Jin, Sheng and Xu, Lumin and Xu, Jin and Wang, Can and Liu, Wentao and Qian, Chen and Ouyang, Wanli and Luo, Ping},
  booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
  year={2020}
}

Results on COCO-WholeBody-Hand val set

Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
pose_hourglass_52	256x256	0.804	0.835	4.54	ckpt	log

Topdown Heatmap + Scnet + Coco + Wholebody on Coco_wholebody_hand¶

@inproceedings{liu2020improving,
  title={Improving Convolutional Networks with Self-Calibrated Convolutions},
  author={Liu, Jiang-Jiang and Hou, Qibin and Cheng, Ming-Ming and Wang, Changhu and Feng, Jiashi},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={10096--10105},
  year={2020}
}

@inproceedings{jin2020whole,
  title={Whole-Body Human Pose Estimation in the Wild},
  author={Jin, Sheng and Xu, Lumin and Xu, Jin and Wang, Can and Liu, Wentao and Qian, Chen and Ouyang, Wanli and Luo, Ping},
  booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
  year={2020}
}

Results on COCO-WholeBody-Hand val set

Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
pose_scnet_50	256x256	0.803	0.834	4.55	ckpt	log

Topdown Heatmap + Resnet + Coco + Wholebody on Coco_wholebody_hand¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

@inproceedings{jin2020whole,
  title={Whole-Body Human Pose Estimation in the Wild},
  author={Jin, Sheng and Xu, Lumin and Xu, Jin and Wang, Can and Liu, Wentao and Qian, Chen and Ouyang, Wanli and Luo, Ping},
  booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
  year={2020}
}

Results on COCO-WholeBody-Hand val set

Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
pose_resnet_50	256x256	0.800	0.833	4.64	ckpt	log

Topdown Heatmap + Hrnetv2 + Coco + Wholebody on Coco_wholebody_hand¶

@article{WangSCJDZLMTWLX19,
  title={Deep High-Resolution Representation Learning for Visual Recognition},
  author={Jingdong Wang and Ke Sun and Tianheng Cheng and
          Borui Jiang and Chaorui Deng and Yang Zhao and Dong Liu and Yadong Mu and
          Mingkui Tan and Xinggang Wang and Wenyu Liu and Bin Xiao},
  journal={TPAMI},
  year={2019}
}

@inproceedings{jin2020whole,
  title={Whole-Body Human Pose Estimation in the Wild},
  author={Jin, Sheng and Xu, Lumin and Xu, Jin and Wang, Can and Liu, Wentao and Qian, Chen and Ouyang, Wanli and Luo, Ping},
  booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
  year={2020}
}

Results on COCO-WholeBody-Hand val set

Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
pose_hrnetv2_w18	256x256	0.813	0.840	4.39	ckpt	log

COFW (ICCV’2013)¶

Topdown Heatmap + Hrnetv2 on Cofw¶

@article{WangSCJDZLMTWLX19,
  title={Deep High-Resolution Representation Learning for Visual Recognition},
  author={Jingdong Wang and Ke Sun and Tianheng Cheng and
          Borui Jiang and Chaorui Deng and Yang Zhao and Dong Liu and Yadong Mu and
          Mingkui Tan and Xinggang Wang and Wenyu Liu and Bin Xiao},
  journal={TPAMI},
  year={2019}
}

COFW (ICCV'2013)

@inproceedings{burgos2013robust,
  title={Robust face landmark estimation under occlusion},
  author={Burgos-Artizzu, Xavier P and Perona, Pietro and Doll{\'a}r, Piotr},
  booktitle={Proceedings of the IEEE international conference on computer vision},
  pages={1513--1520},
  year={2013}
}

Results on COFW dataset

The model is trained on COFW train.

Arch	Input Size	NME	ckpt	log
pose_hrnetv2_w18	256x256	3.40	ckpt	log

DeepFashion (CVPR’2016)¶

Deeppose + Resnet on Deepfashion¶

@inproceedings{toshev2014deeppose,
  title={Deeppose: Human pose estimation via deep neural networks},
  author={Toshev, Alexander and Szegedy, Christian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={1653--1660},
  year={2014}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

DeepFashion (CVPR'2016)

@inproceedings{liuLQWTcvpr16DeepFashion,
 author = {Liu, Ziwei and Luo, Ping and Qiu, Shi and Wang, Xiaogang and Tang, Xiaoou},
 title = {DeepFashion: Powering Robust Clothes Recognition and Retrieval with Rich Annotations},
 booktitle = {Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
 month = {June},
 year = {2016}
}

DeepFashion (ECCV'2016)

@inproceedings{liuYLWTeccv16FashionLandmark,
 author = {Liu, Ziwei and Yan, Sijie and Luo, Ping and Wang, Xiaogang and Tang, Xiaoou},
 title = {Fashion Landmark Detection in the Wild},
 booktitle = {European Conference on Computer Vision (ECCV)},
 month = {October},
 year = {2016}
 }

Results on DeepFashion val set

Set	Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
upper	deeppose_resnet_50	256x256	0.965	0.535	17.2	ckpt	log
lower	deeppose_resnet_50	256x256	0.971	0.678	11.8	ckpt	log
full	deeppose_resnet_50	256x256	0.983	0.602	14.0	ckpt	log

Topdown Heatmap + Resnet on Deepfashion¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

DeepFashion (CVPR'2016)

@inproceedings{liuLQWTcvpr16DeepFashion,
 author = {Liu, Ziwei and Luo, Ping and Qiu, Shi and Wang, Xiaogang and Tang, Xiaoou},
 title = {DeepFashion: Powering Robust Clothes Recognition and Retrieval with Rich Annotations},
 booktitle = {Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
 month = {June},
 year = {2016}
}

DeepFashion (ECCV'2016)

@inproceedings{liuYLWTeccv16FashionLandmark,
 author = {Liu, Ziwei and Yan, Sijie and Luo, Ping and Wang, Xiaogang and Tang, Xiaoou},
 title = {Fashion Landmark Detection in the Wild},
 booktitle = {European Conference on Computer Vision (ECCV)},
 month = {October},
 year = {2016}
 }

Results on DeepFashion val set

Set	Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
upper	pose_resnet_50	256x256	0.954	0.578	16.8	ckpt	log
lower	pose_resnet_50	256x256	0.965	0.744	10.5	ckpt	log
full	pose_resnet_50	256x256	0.977	0.664	12.7	ckpt	log

COCO-WholeBody (ECCV’2020)¶

Associative Embedding + Hrnet on Coco-Wholebody¶

@inproceedings{newell2017associative,
  title={Associative embedding: End-to-end learning for joint detection and grouping},
  author={Newell, Alejandro and Huang, Zhiao and Deng, Jia},
  booktitle={Advances in neural information processing systems},
  pages={2277--2287},
  year={2017}
}

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

@inproceedings{jin2020whole,
  title={Whole-Body Human Pose Estimation in the Wild},
  author={Jin, Sheng and Xu, Lumin and Xu, Jin and Wang, Can and Liu, Wentao and Qian, Chen and Ouyang, Wanli and Luo, Ping},
  booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
  year={2020}
}

Results on COCO-WholeBody v1.0 val without multi-scale test

Arch	Input Size	Body AP	Body AR	Foot AP	Foot AR	Face AP	Face AR	Hand AP	Hand AR	Whole AP	Whole AR	ckpt	log
HRNet-w32+	512x512	0.551	0.650	0.271	0.451	0.564	0.618	0.159	0.238	0.342	0.453	ckpt	log
HRNet-w48+	512x512	0.592	0.686	0.443	0.595	0.619	0.674	0.347	0.438	0.422	0.532	ckpt	log

Note: + means the model is first pre-trained on original COCO dataset, and then fine-tuned on COCO-WholeBody dataset. We find this will lead to better performance.

Associative Embedding + Higherhrnet on Coco-Wholebody¶

@inproceedings{newell2017associative,
  title={Associative embedding: End-to-end learning for joint detection and grouping},
  author={Newell, Alejandro and Huang, Zhiao and Deng, Jia},
  booktitle={Advances in neural information processing systems},
  pages={2277--2287},
  year={2017}
}

@inproceedings{cheng2020higherhrnet,
  title={HigherHRNet: Scale-Aware Representation Learning for Bottom-Up Human Pose Estimation},
  author={Cheng, Bowen and Xiao, Bin and Wang, Jingdong and Shi, Honghui and Huang, Thomas S and Zhang, Lei},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={5386--5395},
  year={2020}
}

@inproceedings{jin2020whole,
  title={Whole-Body Human Pose Estimation in the Wild},
  author={Jin, Sheng and Xu, Lumin and Xu, Jin and Wang, Can and Liu, Wentao and Qian, Chen and Ouyang, Wanli and Luo, Ping},
  booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
  year={2020}
}

Results on COCO-WholeBody v1.0 val without multi-scale test

Arch	Input Size	Body AP	Body AR	Foot AP	Foot AR	Face AP	Face AR	Hand AP	Hand AR	Whole AP	Whole AR	ckpt	log
HigherHRNet-w32+	512x512	0.590	0.672	0.185	0.335	0.676	0.721	0.212	0.298	0.401	0.493	ckpt	log
HigherHRNet-w48+	512x512	0.630	0.706	0.440	0.573	0.730	0.777	0.389	0.477	0.487	0.574	ckpt	log

Note: + means the model is first pre-trained on original COCO dataset, and then fine-tuned on COCO-WholeBody dataset. We find this will lead to better performance.

Topdown Heatmap + Resnet on Coco-Wholebody¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

@inproceedings{jin2020whole,
  title={Whole-Body Human Pose Estimation in the Wild},
  author={Jin, Sheng and Xu, Lumin and Xu, Jin and Wang, Can and Liu, Wentao and Qian, Chen and Ouyang, Wanli and Luo, Ping},
  booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
  year={2020}
}

Results on COCO-WholeBody v1.0 val with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	Body AP	Body AR	Foot AP	Foot AR	Face AP	Face AR	Hand AP	Hand AR	Whole AP	Whole AR	ckpt	log
pose_resnet_50	256x192	0.652	0.739	0.614	0.746	0.608	0.716	0.460	0.584	0.520	0.633	ckpt	log
pose_resnet_50	384x288	0.666	0.747	0.635	0.763	0.732	0.812	0.537	0.647	0.573	0.671	ckpt	log
pose_resnet_101	256x192	0.670	0.754	0.640	0.767	0.611	0.723	0.463	0.589	0.533	0.647	ckpt	log
pose_resnet_101	384x288	0.692	0.770	0.680	0.798	0.747	0.822	0.549	0.658	0.597	0.692	ckpt	log
pose_resnet_152	256x192	0.682	0.764	0.662	0.788	0.624	0.728	0.482	0.606	0.548	0.661	ckpt	log
pose_resnet_152	384x288	0.703	0.780	0.693	0.813	0.751	0.825	0.559	0.667	0.610	0.705	ckpt	log

Topdown Heatmap + Hrnet on Coco-Wholebody¶

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

@inproceedings{jin2020whole,
  title={Whole-Body Human Pose Estimation in the Wild},
  author={Jin, Sheng and Xu, Lumin and Xu, Jin and Wang, Can and Liu, Wentao and Qian, Chen and Ouyang, Wanli and Luo, Ping},
  booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
  year={2020}
}

Results on COCO-WholeBody v1.0 val with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	Body AP	Body AR	Foot AP	Foot AR	Face AP	Face AR	Hand AP	Hand AR	Whole AP	Whole AR	ckpt	log
pose_hrnet_w32	256x192	0.700	0.746	0.567	0.645	0.637	0.688	0.473	0.546	0.553	0.626	ckpt	log
pose_hrnet_w32	384x288	0.701	0.773	0.586	0.692	0.727	0.783	0.516	0.604	0.586	0.674	ckpt	log
pose_hrnet_w48	256x192	0.700	0.776	0.672	0.785	0.656	0.743	0.534	0.639	0.579	0.681	ckpt	log
pose_hrnet_w48	384x288	0.722	0.790	0.694	0.799	0.777	0.834	0.587	0.679	0.631	0.716	ckpt	log

Topdown Heatmap + Tcformer on Coco-Wholebody¶

TCFormer (CVPR'2022)

@inproceedings{zeng2022not,
  title={Not All Tokens Are Equal: Human-centric Visual Analysis via Token Clustering Transformer},
  author={Zeng, Wang and Jin, Sheng and Liu, Wentao and Qian, Chen and Luo, Ping and Ouyang, Wanli and Wang, Xiaogang},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={11101--11111},
  year={2022}
}

@inproceedings{jin2020whole,
  title={Whole-Body Human Pose Estimation in the Wild},
  author={Jin, Sheng and Xu, Lumin and Xu, Jin and Wang, Can and Liu, Wentao and Qian, Chen and Ouyang, Wanli and Luo, Ping},
  booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
  year={2020}
}

Results on COCO-WholeBody v1.0 val with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	Body AP	Body AR	Foot AP	Foot AR	Face AP	Face AR	Hand AP	Hand AR	Whole AP	Whole AR	ckpt	log
tcformer	256x192	0.691	0.769	0.690	0.809	0.650	0.747	0.534	0.647	0.574	0.678	ckpt	log

Topdown Heatmap + Vipnas on Coco-Wholebody¶

ViPNAS (CVPR'2021)

@article{xu2021vipnas,
  title={ViPNAS: Efficient Video Pose Estimation via Neural Architecture Search},
  author={Xu, Lumin and Guan, Yingda and Jin, Sheng and Liu, Wentao and Qian, Chen and Luo, Ping and Ouyang, Wanli and Wang, Xiaogang},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  year={2021}
}

@inproceedings{jin2020whole,
  title={Whole-Body Human Pose Estimation in the Wild},
  author={Jin, Sheng and Xu, Lumin and Xu, Jin and Wang, Can and Liu, Wentao and Qian, Chen and Ouyang, Wanli and Luo, Ping},
  booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
  year={2020}
}

Results on COCO-WholeBody v1.0 val with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	Body AP	Body AR	Foot AP	Foot AR	Face AP	Face AR	Hand AP	Hand AR	Whole AP	Whole AR	ckpt	log
S-ViPNAS-MobileNetV3	256x192	0.619	0.700	0.477	0.608	0.585	0.689	0.386	0.505	0.473	0.578	ckpt	log
S-ViPNAS-Res50	256x192	0.643	0.726	0.553	0.694	0.587	0.698	0.410	0.529	0.495	0.607	ckpt	log

Topdown Heatmap + Vipnas + Dark on Coco-Wholebody¶

ViPNAS (CVPR'2021)

@article{xu2021vipnas,
  title={ViPNAS: Efficient Video Pose Estimation via Neural Architecture Search},
  author={Xu, Lumin and Guan, Yingda and Jin, Sheng and Liu, Wentao and Qian, Chen and Luo, Ping and Ouyang, Wanli and Wang, Xiaogang},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  year={2021}
}

@inproceedings{zhang2020distribution,
  title={Distribution-aware coordinate representation for human pose estimation},
  author={Zhang, Feng and Zhu, Xiatian and Dai, Hanbin and Ye, Mao and Zhu, Ce},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={7093--7102},
  year={2020}
}

@inproceedings{jin2020whole,
  title={Whole-Body Human Pose Estimation in the Wild},
  author={Jin, Sheng and Xu, Lumin and Xu, Jin and Wang, Can and Liu, Wentao and Qian, Chen and Ouyang, Wanli and Luo, Ping},
  booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
  year={2020}
}

Results on COCO-WholeBody v1.0 val with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	Body AP	Body AR	Foot AP	Foot AR	Face AP	Face AR	Hand AP	Hand AR	Whole AP	Whole AR	ckpt	log
S-ViPNAS-MobileNetV3_dark	256x192	0.632	0.710	0.530	0.660	0.672	0.771	0.404	0.519	0.508	0.607	ckpt	log
S-ViPNAS-Res50_dark	256x192	0.650	0.732	0.550	0.686	0.684	0.784	0.437	0.554	0.528	0.632	ckpt	log

Topdown Heatmap + Hrnet + Dark on Coco-Wholebody¶

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

@inproceedings{zhang2020distribution,
  title={Distribution-aware coordinate representation for human pose estimation},
  author={Zhang, Feng and Zhu, Xiatian and Dai, Hanbin and Ye, Mao and Zhu, Ce},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={7093--7102},
  year={2020}
}

@inproceedings{jin2020whole,
  title={Whole-Body Human Pose Estimation in the Wild},
  author={Jin, Sheng and Xu, Lumin and Xu, Jin and Wang, Can and Liu, Wentao and Qian, Chen and Ouyang, Wanli and Luo, Ping},
  booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
  year={2020}
}

Results on COCO-WholeBody v1.0 val with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	Body AP	Body AR	Foot AP	Foot AR	Face AP	Face AR	Hand AP	Hand AR	Whole AP	Whole AR	ckpt	log
pose_hrnet_w32_dark	256x192	0.694	0.764	0.565	0.674	0.736	0.808	0.503	0.602	0.582	0.671	ckpt	log
pose_hrnet_w48_dark+	384x288	0.742	0.807	0.705	0.804	0.840	0.892	0.602	0.694	0.661	0.743	ckpt	log

Note: + means the model is first pre-trained on original COCO dataset, and then fine-tuned on COCO-WholeBody dataset. We find this will lead to better performance.

Horse-10 (WACV’2021)¶

Topdown Heatmap + Resnet on Horse10¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

Horse-10 (WACV'2021)

@inproceedings{mathis2021pretraining,
  title={Pretraining boosts out-of-domain robustness for pose estimation},
  author={Mathis, Alexander and Biasi, Thomas and Schneider, Steffen and Yuksekgonul, Mert and Rogers, Byron and Bethge, Matthias and Mathis, Mackenzie W},
  booktitle={Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision},
  pages={1859--1868},
  year={2021}
}

Results on Horse-10 test set

Set	Arch	Input Size	PCK@0.3	NME	ckpt	log
split1	pose_resnet_50	256x256	0.956	0.113	ckpt	log
split2	pose_resnet_50	256x256	0.954	0.111	ckpt	log
split3	pose_resnet_50	256x256	0.946	0.129	ckpt	log
split1	pose_resnet_101	256x256	0.958	0.115	ckpt	log
split2	pose_resnet_101	256x256	0.955	0.115	ckpt	log
split3	pose_resnet_101	256x256	0.946	0.126	ckpt	log
split1	pose_resnet_152	256x256	0.969	0.105	ckpt	log
split2	pose_resnet_152	256x256	0.970	0.103	ckpt	log
split3	pose_resnet_152	256x256	0.957	0.131	ckpt	log

Topdown Heatmap + Hrnet on Horse10¶

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

Horse-10 (WACV'2021)

@inproceedings{mathis2021pretraining,
  title={Pretraining boosts out-of-domain robustness for pose estimation},
  author={Mathis, Alexander and Biasi, Thomas and Schneider, Steffen and Yuksekgonul, Mert and Rogers, Byron and Bethge, Matthias and Mathis, Mackenzie W},
  booktitle={Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision},
  pages={1859--1868},
  year={2021}
}

Results on Horse-10 test set

Set	Arch	Input Size	PCK@0.3	NME	ckpt	log
split1	pose_hrnet_w32	256x256	0.951	0.122	ckpt	log
split2	pose_hrnet_w32	256x256	0.949	0.116	ckpt	log
split3	pose_hrnet_w32	256x256	0.939	0.153	ckpt	log
split1	pose_hrnet_w48	256x256	0.973	0.095	ckpt	log
split2	pose_hrnet_w48	256x256	0.969	0.101	ckpt	log
split3	pose_hrnet_w48	256x256	0.961	0.128	ckpt	log

Grévy’s Zebra (Elife’2019)¶

Topdown Heatmap + Resnet on Zebra¶

Grévy’s Zebra (Elife'2019)

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

@article{graving2019deepposekit,
  title={DeepPoseKit, a software toolkit for fast and robust animal pose estimation using deep learning},
  author={Graving, Jacob M and Chae, Daniel and Naik, Hemal and Li, Liang and Koger, Benjamin and Costelloe, Blair R and Couzin, Iain D},
  journal={Elife},
  volume={8},
  pages={e47994},
  year={2019},
  publisher={eLife Sciences Publications Limited}
}

Results on Grévy’s Zebra test set

Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
pose_resnet_50	160x160	1.000	0.914	1.86	ckpt	log
pose_resnet_101	160x160	1.000	0.916	1.82	ckpt	log
pose_resnet_152	160x160	1.000	0.921	1.66	ckpt	log

COCO (ECCV’2014)¶

Associative Embedding + Higherhrnet + Udp on Coco¶

@inproceedings{newell2017associative,
  title={Associative embedding: End-to-end learning for joint detection and grouping},
  author={Newell, Alejandro and Huang, Zhiao and Deng, Jia},
  booktitle={Advances in neural information processing systems},
  pages={2277--2287},
  year={2017}
}

@inproceedings{cheng2020higherhrnet,
  title={HigherHRNet: Scale-Aware Representation Learning for Bottom-Up Human Pose Estimation},
  author={Cheng, Bowen and Xiao, Bin and Wang, Jingdong and Shi, Honghui and Huang, Thomas S and Zhang, Lei},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={5386--5395},
  year={2020}
}

@InProceedings{Huang_2020_CVPR,
  author = {Huang, Junjie and Zhu, Zheng and Guo, Feng and Huang, Guan},
  title = {The Devil Is in the Details: Delving Into Unbiased Data Processing for Human Pose Estimation},
  booktitle = {The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  month = {June},
  year = {2020}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 without multi-scale test

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
HigherHRNet-w32_udp	512x512	0.678	0.862	0.736	0.724	0.890	ckpt	log
HigherHRNet-w48_udp	512x512	0.690	0.872	0.750	0.734	0.891	ckpt	log

Associative Embedding + Mobilenetv2 on Coco¶

@inproceedings{newell2017associative,
  title={Associative embedding: End-to-end learning for joint detection and grouping},
  author={Newell, Alejandro and Huang, Zhiao and Deng, Jia},
  booktitle={Advances in neural information processing systems},
  pages={2277--2287},
  year={2017}
}

@inproceedings{sandler2018mobilenetv2,
  title={Mobilenetv2: Inverted residuals and linear bottlenecks},
  author={Sandler, Mark and Howard, Andrew and Zhu, Menglong and Zhmoginov, Andrey and Chen, Liang-Chieh},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={4510--4520},
  year={2018}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 without multi-scale test

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_mobilenetv2	512x512	0.380	0.671	0.368	0.473	0.741	ckpt	log

Results on COCO val2017 with multi-scale test. 3 default scales ([2, 1, 0.5]) are used

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_mobilenetv2	512x512	0.442	0.696	0.422	0.517	0.766	ckpt	log

Associative Embedding + Hrnet on Coco¶

@inproceedings{newell2017associative,
  title={Associative embedding: End-to-end learning for joint detection and grouping},
  author={Newell, Alejandro and Huang, Zhiao and Deng, Jia},
  booktitle={Advances in neural information processing systems},
  pages={2277--2287},
  year={2017}
}

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 without multi-scale test

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
HRNet-w32	512x512	0.654	0.863	0.720	0.710	0.892	ckpt	log
HRNet-w48	512x512	0.665	0.860	0.727	0.716	0.889	ckpt	log

Results on COCO val2017 with multi-scale test. 3 default scales ([2, 1, 0.5]) are used

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
HRNet-w32	512x512	0.698	0.877	0.760	0.748	0.907	ckpt	log
HRNet-w48	512x512	0.712	0.880	0.771	0.757	0.909	ckpt	log

Associative Embedding + Higherhrnet on Coco¶

@inproceedings{newell2017associative,
  title={Associative embedding: End-to-end learning for joint detection and grouping},
  author={Newell, Alejandro and Huang, Zhiao and Deng, Jia},
  booktitle={Advances in neural information processing systems},
  pages={2277--2287},
  year={2017}
}

@inproceedings{cheng2020higherhrnet,
  title={HigherHRNet: Scale-Aware Representation Learning for Bottom-Up Human Pose Estimation},
  author={Cheng, Bowen and Xiao, Bin and Wang, Jingdong and Shi, Honghui and Huang, Thomas S and Zhang, Lei},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={5386--5395},
  year={2020}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 without multi-scale test

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
HigherHRNet-w32	512x512	0.677	0.870	0.738	0.723	0.890	ckpt	log
HigherHRNet-w32	640x640	0.686	0.871	0.747	0.733	0.898	ckpt	log
HigherHRNet-w48	512x512	0.686	0.873	0.741	0.731	0.892	ckpt	log

Results on COCO val2017 with multi-scale test. 3 default scales ([2, 1, 0.5]) are used

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
HigherHRNet-w32	512x512	0.706	0.881	0.771	0.747	0.901	ckpt	log
HigherHRNet-w32	640x640	0.706	0.880	0.770	0.749	0.902	ckpt	log
HigherHRNet-w48	512x512	0.716	0.884	0.775	0.755	0.901	ckpt	log

Associative Embedding + Resnet on Coco¶

@inproceedings{newell2017associative,
  title={Associative embedding: End-to-end learning for joint detection and grouping},
  author={Newell, Alejandro and Huang, Zhiao and Deng, Jia},
  booktitle={Advances in neural information processing systems},
  pages={2277--2287},
  year={2017}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 without multi-scale test

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_resnet_50	512x512	0.466	0.742	0.479	0.552	0.797	ckpt	log
pose_resnet_50	640x640	0.479	0.757	0.487	0.566	0.810	ckpt	log
pose_resnet_101	512x512	0.554	0.807	0.599	0.622	0.841	ckpt	log
pose_resnet_152	512x512	0.595	0.829	0.648	0.651	0.856	ckpt	log

Results on COCO val2017 with multi-scale test. 3 default scales ([2, 1, 0.5]) are used

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_resnet_50	512x512	0.503	0.765	0.521	0.591	0.821	ckpt	log
pose_resnet_50	640x640	0.525	0.784	0.542	0.610	0.832	ckpt	log
pose_resnet_101	512x512	0.603	0.831	0.641	0.668	0.870	ckpt	log
pose_resnet_152	512x512	0.660	0.860	0.713	0.709	0.889	ckpt	log

Associative Embedding + Hrnet + Udp on Coco¶

@inproceedings{newell2017associative,
  title={Associative embedding: End-to-end learning for joint detection and grouping},
  author={Newell, Alejandro and Huang, Zhiao and Deng, Jia},
  booktitle={Advances in neural information processing systems},
  pages={2277--2287},
  year={2017}
}

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

@InProceedings{Huang_2020_CVPR,
  author = {Huang, Junjie and Zhu, Zheng and Guo, Feng and Huang, Guan},
  title = {The Devil Is in the Details: Delving Into Unbiased Data Processing for Human Pose Estimation},
  booktitle = {The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  month = {June},
  year = {2020}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 without multi-scale test

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
HRNet-w32_udp	512x512	0.671	0.863	0.729	0.717	0.889	ckpt	log
HRNet-w48_udp	512x512	0.681	0.872	0.741	0.725	0.892	ckpt	log

Associative Embedding + Hourglass + Ae on Coco¶

HourglassAENet (NIPS'2017)

@inproceedings{newell2017associative,
  title={Associative embedding: End-to-end learning for joint detection and grouping},
  author={Newell, Alejandro and Huang, Zhiao and Deng, Jia},
  booktitle={Advances in neural information processing systems},
  pages={2277--2287},
  year={2017}
}

@inproceedings{newell2017associative,
  title={Associative embedding: End-to-end learning for joint detection and grouping},
  author={Newell, Alejandro and Huang, Zhiao and Deng, Jia},
  booktitle={Advances in neural information processing systems},
  pages={2277--2287},
  year={2017}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 without multi-scale test

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_hourglass_ae	512x512	0.613	0.833	0.667	0.659	0.850	ckpt	log

Results on COCO val2017 with multi-scale test. 3 default scales ([2, 1, 0.5]) are used

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_hourglass_ae	512x512	0.667	0.855	0.723	0.707	0.877	ckpt	log

Deeppose + Resnet on Coco¶

@inproceedings{toshev2014deeppose,
  title={Deeppose: Human pose estimation via deep neural networks},
  author={Toshev, Alexander and Szegedy, Christian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={1653--1660},
  year={2014}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
deeppose_resnet_50	256x192	0.526	0.816	0.586	0.638	0.887	ckpt	log
deeppose_resnet_101	256x192	0.560	0.832	0.628	0.668	0.900	ckpt	log
deeppose_resnet_152	256x192	0.583	0.843	0.659	0.686	0.907	ckpt	log

Deeppose + Resnet + Rle on Coco¶

@inproceedings{toshev2014deeppose,
  title={Deeppose: Human pose estimation via deep neural networks},
  author={Toshev, Alexander and Szegedy, Christian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={1653--1660},
  year={2014}
}

RLE (ICCV'2021)

@inproceedings{li2021human,
  title={Human pose regression with residual log-likelihood estimation},
  author={Li, Jiefeng and Bian, Siyuan and Zeng, Ailing and Wang, Can and Pang, Bo and Liu, Wentao and Lu, Cewu},
  booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
  pages={11025--11034},
  year={2021}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
deeppose_resnet_50_rle	256x192	0.704	0.883	0.777	0.751	0.920	ckpt	log
deeppose_resnet_101_rle	256x192	0.722	0.894	0.794	0.768	0.930	ckpt	log
deeppose_resnet_152_rle	256x192	0.731	0.897	0.805	0.777	0.933	ckpt	log
deeppose_resnet_152_rle	384x288	0.749	0.901	0.815	0.793	0.935	ckpt	log

Dekr + Hrnet on Coco¶

DEKR (CVPR'2021)

@inproceedings{geng2021bottom,
  title={Bottom-up human pose estimation via disentangled keypoint regression},
  author={Geng, Zigang and Sun, Ke and Xiao, Bin and Zhang, Zhaoxiang and Wang, Jingdong},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={14676--14686},
  year={2021}
}

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 without multi-scale test

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
HRNet-w32	512x512	0.680	0.868	0.745	0.728	0.897	ckpt	log
HRNet-w48	640x640	0.709	0.876	0.773	0.758	0.909	ckpt	log

Results on COCO val2017 with multi-scale test. 3 default scales ([2, 1, 0.5]) are used

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt
HRNet-w32*	512x512	0.705	0.878	0.767	0.759	0.921	ckpt
HRNet-w48*	640x640	0.722	0.882	0.785	0.778	0.928	ckpt

* these configs are generally used for evaluation. The training settings are identical to their single-scale counterparts.

The results of models provided by the authors on COCO val2017 using the same evaluation protocol

Arch	Input Size	Setting	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt
HRNet-w32	512x512	single-scale	0.678	0.868	0.744	0.728	0.897	see official implementation
HRNet-w48	640x640	single-scale	0.707	0.876	0.773	0.757	0.909	see official implementation
HRNet-w32	512x512	multi-scale	0.708	0.880	0.773	0.763	0.921	see official implementation
HRNet-w48	640x640	multi-scale	0.721	0.881	0.786	0.779	0.927	see official implementation

The discrepancy between these results and that shown in paper is attributed to the differences in implementation details in evaluation process.

Topdown Heatmap + Seresnet on Coco¶

SEResNet (CVPR'2018)

@inproceedings{hu2018squeeze,
  title={Squeeze-and-excitation networks},
  author={Hu, Jie and Shen, Li and Sun, Gang},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={7132--7141},
  year={2018}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_seresnet_50	256x192	0.728	0.900	0.809	0.784	0.940	ckpt	log
pose_seresnet_50	384x288	0.748	0.905	0.819	0.799	0.941	ckpt	log
pose_seresnet_101	256x192	0.734	0.904	0.815	0.790	0.942	ckpt	log
pose_seresnet_101	384x288	0.753	0.907	0.823	0.805	0.943	ckpt	log
pose_seresnet_152*	256x192	0.730	0.899	0.810	0.786	0.940	ckpt	log
pose_seresnet_152*	384x288	0.753	0.906	0.823	0.806	0.945	ckpt	log

Note that * means without imagenet pre-training.

Topdown Heatmap + Resnetv1d on Coco¶

ResNetV1D (CVPR'2019)

@inproceedings{he2019bag,
  title={Bag of tricks for image classification with convolutional neural networks},
  author={He, Tong and Zhang, Zhi and Zhang, Hang and Zhang, Zhongyue and Xie, Junyuan and Li, Mu},
  booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
  pages={558--567},
  year={2019}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_resnetv1d_50	256x192	0.722	0.897	0.799	0.777	0.933	ckpt	log
pose_resnetv1d_50	384x288	0.730	0.900	0.799	0.780	0.934	ckpt	log
pose_resnetv1d_101	256x192	0.731	0.899	0.809	0.786	0.938	ckpt	log
pose_resnetv1d_101	384x288	0.748	0.902	0.816	0.799	0.939	ckpt	log
pose_resnetv1d_152	256x192	0.737	0.902	0.812	0.791	0.940	ckpt	log
pose_resnetv1d_152	384x288	0.752	0.909	0.821	0.802	0.944	ckpt	log

Topdown Heatmap + Hourglass on Coco¶

@inproceedings{newell2016stacked,
  title={Stacked hourglass networks for human pose estimation},
  author={Newell, Alejandro and Yang, Kaiyu and Deng, Jia},
  booktitle={European conference on computer vision},
  pages={483--499},
  year={2016},
  organization={Springer}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_hourglass_52	256x256	0.726	0.896	0.799	0.780	0.934	ckpt	log
pose_hourglass_52	384x384	0.746	0.900	0.813	0.797	0.939	ckpt	log

Topdown Heatmap + RSN on Coco¶

RSN (ECCV'2020)

@misc{cai2020learning,
    title={Learning Delicate Local Representations for Multi-Person Pose Estimation},
    author={Yuanhao Cai and Zhicheng Wang and Zhengxiong Luo and Binyi Yin and Angang Du and Haoqian Wang and Xinyu Zhou and Erjin Zhou and Xiangyu Zhang and Jian Sun},
    year={2020},
    eprint={2003.04030},
    archivePrefix={arXiv},
    primaryClass={cs.CV}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
rsn_18	256x192	0.704	0.887	0.779	0.771	0.926	ckpt	log
rsn_50	256x192	0.723	0.896	0.800	0.788	0.934	ckpt	log
2xrsn_50	256x192	0.745	0.899	0.818	0.809	0.939	ckpt	log
3xrsn_50	256x192	0.750	0.900	0.823	0.813	0.940	ckpt	log

Topdown Heatmap + Resnet + Fp16 on Coco¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

FP16 (ArXiv'2017)

@article{micikevicius2017mixed,
  title={Mixed precision training},
  author={Micikevicius, Paulius and Narang, Sharan and Alben, Jonah and Diamos, Gregory and Elsen, Erich and Garcia, David and Ginsburg, Boris and Houston, Michael and Kuchaiev, Oleksii and Venkatesh, Ganesh and others},
  journal={arXiv preprint arXiv:1710.03740},
  year={2017}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_resnet_50_fp16	256x192	0.717	0.898	0.793	0.772	0.936	ckpt	log

Topdown Heatmap + Mobilenetv2 on Coco¶

@inproceedings{sandler2018mobilenetv2,
  title={Mobilenetv2: Inverted residuals and linear bottlenecks},
  author={Sandler, Mark and Howard, Andrew and Zhu, Menglong and Zhmoginov, Andrey and Chen, Liang-Chieh},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={4510--4520},
  year={2018}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_mobilenetv2	256x192	0.646	0.874	0.723	0.707	0.917	ckpt	log
pose_mobilenetv2	384x288	0.673	0.879	0.743	0.729	0.916	ckpt	log

Topdown Heatmap + Shufflenetv1 on Coco¶

ShufflenetV1 (CVPR'2018)

@inproceedings{zhang2018shufflenet,
  title={Shufflenet: An extremely efficient convolutional neural network for mobile devices},
  author={Zhang, Xiangyu and Zhou, Xinyu and Lin, Mengxiao and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={6848--6856},
  year={2018}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_shufflenetv1	256x192	0.585	0.845	0.650	0.651	0.894	ckpt	log
pose_shufflenetv1	384x288	0.622	0.859	0.685	0.684	0.901	ckpt	log

Topdown Heatmap + MSPN on Coco¶

MSPN (ArXiv'2019)

@article{li2019rethinking,
  title={Rethinking on Multi-Stage Networks for Human Pose Estimation},
  author={Li, Wenbo and Wang, Zhicheng and Yin, Binyi and Peng, Qixiang and Du, Yuming and Xiao, Tianzi and Yu, Gang and Lu, Hongtao and Wei, Yichen and Sun, Jian},
  journal={arXiv preprint arXiv:1901.00148},
  year={2019}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
mspn_50	256x192	0.723	0.895	0.794	0.788	0.933	ckpt	log
2xmspn_50	256x192	0.754	0.903	0.825	0.815	0.941	ckpt	log
3xmspn_50	256x192	0.758	0.904	0.830	0.821	0.943	ckpt	log
4xmspn_50	256x192	0.764	0.906	0.835	0.826	0.944	ckpt	log

Topdown Heatmap + Hrnet + Fp16 on Coco¶

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

FP16 (ArXiv'2017)

@article{micikevicius2017mixed,
  title={Mixed precision training},
  author={Micikevicius, Paulius and Narang, Sharan and Alben, Jonah and Diamos, Gregory and Elsen, Erich and Garcia, David and Ginsburg, Boris and Houston, Michael and Kuchaiev, Oleksii and Venkatesh, Ganesh and others},
  journal={arXiv preprint arXiv:1710.03740},
  year={2017}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_hrnet_w32_fp16	256x192	0.746	0.905	0.88	0.800	0.943	ckpt	log

Topdown Heatmap + Hrnet on Coco¶

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_hrnet_w32	256x192	0.746	0.904	0.819	0.799	0.942	ckpt	log
pose_hrnet_w32	384x288	0.760	0.906	0.829	0.810	0.943	ckpt	log
pose_hrnet_w48	256x192	0.756	0.907	0.825	0.806	0.942	ckpt	log
pose_hrnet_w48	384x288	0.767	0.910	0.831	0.816	0.946	ckpt	log

Topdown Heatmap + Resnext on Coco¶

ResNext (CVPR'2017)

@inproceedings{xie2017aggregated,
  title={Aggregated residual transformations for deep neural networks},
  author={Xie, Saining and Girshick, Ross and Doll{\'a}r, Piotr and Tu, Zhuowen and He, Kaiming},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={1492--1500},
  year={2017}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_resnext_50	256x192	0.714	0.898	0.789	0.771	0.937	ckpt	log
pose_resnext_50	384x288	0.724	0.899	0.794	0.777	0.935	ckpt	log
pose_resnext_101	256x192	0.726	0.900	0.801	0.782	0.940	ckpt	log
pose_resnext_101	384x288	0.743	0.903	0.815	0.795	0.939	ckpt	log
pose_resnext_152	256x192	0.730	0.904	0.808	0.786	0.940	ckpt	log
pose_resnext_152	384x288	0.742	0.902	0.810	0.794	0.939	ckpt	log

Topdown Heatmap + Resnet + Dark on Coco¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

@inproceedings{zhang2020distribution,
  title={Distribution-aware coordinate representation for human pose estimation},
  author={Zhang, Feng and Zhu, Xiatian and Dai, Hanbin and Ye, Mao and Zhu, Ce},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={7093--7102},
  year={2020}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_resnet_50_dark	256x192	0.724	0.898	0.800	0.777	0.936	ckpt	log
pose_resnet_50_dark	384x288	0.735	0.900	0.801	0.785	0.937	ckpt	log
pose_resnet_101_dark	256x192	0.732	0.899	0.808	0.786	0.938	ckpt	log
pose_resnet_101_dark	384x288	0.749	0.902	0.816	0.799	0.939	ckpt	log
pose_resnet_152_dark	256x192	0.745	0.905	0.821	0.797	0.942	ckpt	log
pose_resnet_152_dark	384x288	0.757	0.909	0.826	0.806	0.943	ckpt	log

Topdown Heatmap + VGG on Coco¶

VGG (ICLR'2015)

@article{simonyan2014very,
  title={Very deep convolutional networks for large-scale image recognition},
  author={Simonyan, Karen and Zisserman, Andrew},
  journal={arXiv preprint arXiv:1409.1556},
  year={2014}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
vgg	256x192	0.698	0.890	0.768	0.754	0.929	ckpt	log

Topdown Heatmap + Shufflenetv2 on Coco¶

ShufflenetV2 (ECCV'2018)

@inproceedings{ma2018shufflenet,
  title={Shufflenet v2: Practical guidelines for efficient cnn architecture design},
  author={Ma, Ningning and Zhang, Xiangyu and Zheng, Hai-Tao and Sun, Jian},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={116--131},
  year={2018}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_shufflenetv2	256x192	0.599	0.854	0.663	0.664	0.899	ckpt	log
pose_shufflenetv2	384x288	0.636	0.865	0.705	0.697	0.909	ckpt	log

Topdown Heatmap + Hrnet + Augmentation on Coco¶

Albumentations (Information'2020)

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

@article{buslaev2020albumentations,
  title={Albumentations: fast and flexible image augmentations},
  author={Buslaev, Alexander and Iglovikov, Vladimir I and Khvedchenya, Eugene and Parinov, Alex and Druzhinin, Mikhail and Kalinin, Alexandr A},
  journal={Information},
  volume={11},
  number={2},
  pages={125},
  year={2020},
  publisher={Multidisciplinary Digital Publishing Institute}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
coarsedropout	256x192	0.753	0.908	0.822	0.806	0.946	ckpt	log
gridmask	256x192	0.752	0.906	0.825	0.804	0.943	ckpt	log
photometric	256x192	0.753	0.909	0.825	0.805	0.943	ckpt	log

Topdown Heatmap + Swin on Coco¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

Swin (ICCV'2021)

@inproceedings{liu2021swin,
  title={Swin transformer: Hierarchical vision transformer using shifted windows},
  author={Liu, Ze and Lin, Yutong and Cao, Yue and Hu, Han and Wei, Yixuan and Zhang, Zheng and Lin, Stephen and Guo, Baining},
  booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
  pages={10012--10022},
  year={2021}
}

FPN (CVPR'2017)

@inproceedings{lin2017feature,
  title={Feature pyramid networks for object detection},
  author={Lin, Tsung-Yi and Doll{\'a}r, Piotr and Girshick, Ross and He, Kaiming and Hariharan, Bharath and Belongie, Serge},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={2117--2125},
  year={2017}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_swin_t	256x192	0.724	0.901	0.806	0.782	0.940	ckpt	log
pose_swin_b	256x192	0.737	0.904	0.820	0.798	0.946	ckpt	log
pose_swin_b	384x288	0.759	0.910	0.832	0.811	0.946	ckpt	log
pose_swin_l	256x192	0.743	0.906	0.821	0.798	0.943	ckpt	log
pose_swin_l	384x288	0.763	0.912	0.830	0.814	0.949	ckpt	log
pose_swin_b_fpn	256x192	0.741	0.907	0.821	0.798	0.946	ckpt	log

Topdown Heatmap + Litehrnet on Coco¶

LiteHRNet (CVPR'2021)

@inproceedings{Yulitehrnet21,
  title={Lite-HRNet: A Lightweight High-Resolution Network},
  author={Yu, Changqian and Xiao, Bin and Gao, Changxin and Yuan, Lu and Zhang, Lei and Sang, Nong and Wang, Jingdong},
  booktitle={CVPR},
  year={2021}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
LiteHRNet-18	256x192	0.643	0.868	0.720	0.706	0.912	ckpt	log
LiteHRNet-18	384x288	0.677	0.878	0.746	0.735	0.920	ckpt	log
LiteHRNet-30	256x192	0.675	0.881	0.754	0.736	0.924	ckpt	log
LiteHRNet-30	384x288	0.700	0.884	0.776	0.758	0.928	ckpt	log

Topdown Heatmap + Scnet on Coco¶

@inproceedings{liu2020improving,
  title={Improving Convolutional Networks with Self-Calibrated Convolutions},
  author={Liu, Jiang-Jiang and Hou, Qibin and Cheng, Ming-Ming and Wang, Changhu and Feng, Jiashi},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={10096--10105},
  year={2020}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_scnet_50	256x192	0.728	0.899	0.807	0.784	0.938	ckpt	log
pose_scnet_50	384x288	0.751	0.906	0.818	0.802	0.943	ckpt	log
pose_scnet_101	256x192	0.733	0.903	0.813	0.790	0.941	ckpt	log
pose_scnet_101	384x288	0.752	0.906	0.823	0.804	0.943	ckpt	log

Topdown Heatmap + Hrformer on Coco¶

HRFormer (NIPS'2021)

@article{yuan2021hrformer,
  title={HRFormer: High-Resolution Vision Transformer for Dense Predict},
  author={Yuan, Yuhui and Fu, Rao and Huang, Lang and Lin, Weihong and Zhang, Chao and Chen, Xilin and Wang, Jingdong},
  journal={Advances in Neural Information Processing Systems},
  volume={34},
  year={2021}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_hrformer_small	256x192	0.738	0.904	0.811	0.792	0.941	ckpt	log
pose_hrformer_small	384x288	0.757	0.905	0.824	0.807	0.941	ckpt	log
pose_hrformer_base	256x192	0.753	0.907	0.826	0.807	0.943	ckpt	log
pose_hrformer_base	384x288	0.774	0.909	0.842	0.823	0.945	ckpt	log

Topdown Heatmap + Resnest on Coco¶

ResNeSt (ArXiv'2020)

@article{zhang2020resnest,
  title={ResNeSt: Split-Attention Networks},
  author={Zhang, Hang and Wu, Chongruo and Zhang, Zhongyue and Zhu, Yi and Zhang, Zhi and Lin, Haibin and Sun, Yue and He, Tong and Muller, Jonas and Manmatha, R. and Li, Mu and Smola, Alexander},
  journal={arXiv preprint arXiv:2004.08955},
  year={2020}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_resnest_50	256x192	0.721	0.899	0.802	0.776	0.938	ckpt	log
pose_resnest_50	384x288	0.737	0.900	0.811	0.789	0.938	ckpt	log
pose_resnest_101	256x192	0.725	0.899	0.807	0.781	0.939	ckpt	log
pose_resnest_101	384x288	0.746	0.906	0.820	0.798	0.943	ckpt	log
pose_resnest_200	256x192	0.732	0.905	0.812	0.787	0.942	ckpt	log
pose_resnest_200	384x288	0.754	0.908	0.827	0.807	0.945	ckpt	log
pose_resnest_269	256x192	0.738	0.907	0.819	0.793	0.945	ckpt	log
pose_resnest_269	384x288	0.755	0.908	0.828	0.806	0.943	ckpt	log

Topdown Heatmap + PVT on Coco¶

PVT (ICCV'2021)

@inproceedings{wang2021pyramid,
  title={Pyramid vision transformer: A versatile backbone for dense prediction without convolutions},
  author={Wang, Wenhai and Xie, Enze and Li, Xiang and Fan, Deng-Ping and Song, Kaitao and Liang, Ding and Lu, Tong and Luo, Ping and Shao, Ling},
  booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
  pages={568--578},
  year={2021}
}

PVTV2 (CVMJ'2022)

@article{wang2022pvt,
  title={PVT v2: Improved baselines with Pyramid Vision Transformer},
  author={Wang, Wenhai and Xie, Enze and Li, Xiang and Fan, Deng-Ping and Song, Kaitao and Liang, Ding and Lu, Tong and Luo, Ping and Shao, Ling},
  journal={Computational Visual Media},
  pages={1--10},
  year={2022},
  publisher={Springer}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_pvt-s	256x192	0.714	0.896	0.794	0.773	0.936	ckpt	log
pose_pvtv2-b2	256x192	0.737	0.905	0.812	0.791	0.942	ckpt	log

Topdown Heatmap + Resnet on Coco¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_resnet_50	256x192	0.718	0.898	0.795	0.773	0.937	ckpt	log
pose_resnet_50	384x288	0.731	0.900	0.799	0.783	0.931	ckpt	log
pose_resnet_101	256x192	0.726	0.899	0.806	0.781	0.939	ckpt	log
pose_resnet_101	384x288	0.748	0.905	0.817	0.798	0.940	ckpt	log
pose_resnet_152	256x192	0.735	0.905	0.812	0.790	0.943	ckpt	log
pose_resnet_152	384x288	0.750	0.908	0.821	0.800	0.942	ckpt	log

Topdown Heatmap + Alexnet on Coco¶

AlexNet (NeurIPS'2012)

@inproceedings{krizhevsky2012imagenet,
  title={Imagenet classification with deep convolutional neural networks},
  author={Krizhevsky, Alex and Sutskever, Ilya and Hinton, Geoffrey E},
  booktitle={Advances in neural information processing systems},
  pages={1097--1105},
  year={2012}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_alexnet	256x192	0.397	0.758	0.381	0.478	0.822	ckpt	log

Topdown Heatmap + Hrnet + Udp on Coco¶

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

@InProceedings{Huang_2020_CVPR,
  author = {Huang, Junjie and Zhu, Zheng and Guo, Feng and Huang, Guan},
  title = {The Devil Is in the Details: Delving Into Unbiased Data Processing for Human Pose Estimation},
  booktitle = {The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  month = {June},
  year = {2020}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_hrnet_w32_udp	256x192	0.760	0.907	0.827	0.811	0.945	ckpt	log
pose_hrnet_w32_udp	384x288	0.769	0.908	0.833	0.817	0.944	ckpt	log
pose_hrnet_w48_udp	256x192	0.767	0.906	0.834	0.817	0.945	ckpt	log
pose_hrnet_w48_udp	384x288	0.772	0.910	0.835	0.820	0.945	ckpt	log
pose_hrnet_w32_udp_regress	256x192	0.758	0.908	0.823	0.812	0.943	ckpt	log

Note that, UDP also adopts the unbiased encoding/decoding algorithm of DARK.

Topdown Heatmap + Hrnet + Dark on Coco¶

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

@inproceedings{zhang2020distribution,
  title={Distribution-aware coordinate representation for human pose estimation},
  author={Zhang, Feng and Zhu, Xiatian and Dai, Hanbin and Ye, Mao and Zhu, Ce},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={7093--7102},
  year={2020}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
pose_hrnet_w32_dark	256x192	0.757	0.907	0.823	0.808	0.943	ckpt	log
pose_hrnet_w32_dark	384x288	0.766	0.907	0.831	0.815	0.943	ckpt	log
pose_hrnet_w48_dark	256x192	0.764	0.907	0.830	0.814	0.943	ckpt	log
pose_hrnet_w48_dark	384x288	0.772	0.910	0.836	0.820	0.946	ckpt	log

Topdown Heatmap + Vipnas on Coco¶

ViPNAS (CVPR'2021)

@article{xu2021vipnas,
  title={ViPNAS: Efficient Video Pose Estimation via Neural Architecture Search},
  author={Xu, Lumin and Guan, Yingda and Jin, Sheng and Liu, Wentao and Qian, Chen and Luo, Ping and Ouyang, Wanli and Wang, Xiaogang},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  year={2021}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
S-ViPNAS-MobileNetV3	256x192	0.700	0.887	0.778	0.757	0.929	ckpt	log
S-ViPNAS-Res50	256x192	0.711	0.893	0.789	0.769	0.934	ckpt	log

Topdown Heatmap + CPM on Coco¶

CPM (CVPR'2016)

@inproceedings{wei2016convolutional,
  title={Convolutional pose machines},
  author={Wei, Shih-En and Ramakrishna, Varun and Kanade, Takeo and Sheikh, Yaser},
  booktitle={Proceedings of the IEEE conference on Computer Vision and Pattern Recognition},
  pages={4724--4732},
  year={2016}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 with detector having human AP of 56.4 on COCO val2017 dataset

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
cpm	256x192	0.623	0.859	0.704	0.686	0.903	ckpt	log
cpm	384x288	0.650	0.864	0.725	0.708	0.905	ckpt	log

Cid + Hrnet on Coco¶

CID (CVPR'2022)

@InProceedings{Wang_2022_CVPR,
    author    = {Wang, Dongkai and Zhang, Shiliang},
    title     = {Contextual Instance Decoupling for Robust Multi-Person Pose Estimation},
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
    month     = {June},
    year      = {2022},
    pages     = {11060-11068}
}

PoseWarper (NeurIPS'2019)

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Results on COCO val2017 without multi-scale test

Arch	Input Size	AP	AP⁵⁰	AP⁷⁵	AR	AR⁵⁰	ckpt	log
CID	512x512	0.702	0.887	0.768	0.755	0.926	ckpt	log
CID	512x512	0.715	0.895	0.780	0.768	0.932	ckpt	log

Posewarper + Hrnet + Posetrack18 on Posetrack18¶

@inproceedings{NIPS2019_gberta,
title = {Learning Temporal Pose Estimation from Sparsely Labeled Videos},
author = {Bertasius, Gedas and Feichtenhofer, Christoph, and Tran, Du and Shi, Jianbo, and Torresani, Lorenzo},
booktitle = {Advances in Neural Information Processing Systems 33},
year = {2019},
}

@inproceedings{sun2019deep,
  title={Deep high-resolution representation learning for human pose estimation},
  author={Sun, Ke and Xiao, Bin and Liu, Dong and Wang, Jingdong},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={5693--5703},
  year={2019}
}

@inproceedings{andriluka2018posetrack,
  title={Posetrack: A benchmark for human pose estimation and tracking},
  author={Andriluka, Mykhaylo and Iqbal, Umar and Insafutdinov, Eldar and Pishchulin, Leonid and Milan, Anton and Gall, Juergen and Schiele, Bernt},
  booktitle={Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition},
  pages={5167--5176},
  year={2018}
}

@inproceedings{lin2014microsoft,
  title={Microsoft coco: Common objects in context},
  author={Lin, Tsung-Yi and Maire, Michael and Belongie, Serge and Hays, James and Perona, Pietro and Ramanan, Deva and Doll{\'a}r, Piotr and Zitnick, C Lawrence},
  booktitle={European conference on computer vision},
  pages={740--755},
  year={2014},
  organization={Springer}
}

Note that the training of PoseWarper can be split into two stages.

The first-stage is trained with the pre-trained checkpoint on COCO dataset, and the main backbone is fine-tuned on PoseTrack18 in a single-frame setting.

The second-stage is trained with the last checkpoint from the first stage, and the warping offsets are learned in a multi-frame setting while the backbone is frozen.

Results on PoseTrack2018 val with ground-truth bounding boxes

Arch	Input Size	Head	Shou	Elb	Wri	Hip	Knee	Ankl	Total	ckpt	log
pose_hrnet_w48	384x288	88.2	90.3	86.1	81.6	81.8	83.8	81.5	85.0	ckpt	log

Results on PoseTrack2018 val with precomputed human bounding boxes from PoseWarper supplementary data files from this link¹.

Arch	Input Size	Head	Shou	Elb	Wri	Hip	Knee	Ankl	Total	ckpt	log
pose_hrnet_w48	384x288	81.8	85.6	82.7	77.2	76.8	79.0	74.4	79.8	ckpt	log

¹ Please download the precomputed human bounding boxes on PoseTrack2018 val from $PoseWarper_supp_files/posetrack18_precomputed_boxes/val_boxes.json and place it here: $mmpose/data/posetrack18/posetrack18_precomputed_boxes/val_boxes.json to be consistent with the config. Please refer to DATA Preparation for more detail about data preparation.

RHD (ICCV’2017)¶

Deeppose + Resnet on Rhd2d¶

@inproceedings{toshev2014deeppose,
  title={Deeppose: Human pose estimation via deep neural networks},
  author={Toshev, Alexander and Szegedy, Christian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={1653--1660},
  year={2014}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

@TechReport{zb2017hand,
  author={Christian Zimmermann and Thomas Brox},
  title={Learning to Estimate 3D Hand Pose from Single RGB Images},
  institution={arXiv:1705.01389},
  year={2017},
  note="https://arxiv.org/abs/1705.01389",
  url="https://lmb.informatik.uni-freiburg.de/projects/hand3d/"
}

Results on RHD test set

Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
deeppose_resnet_50	256x256	0.988	0.865	3.29	ckpt	log

Topdown Heatmap + Mobilenetv2 on Rhd2d¶

@inproceedings{sandler2018mobilenetv2,
  title={Mobilenetv2: Inverted residuals and linear bottlenecks},
  author={Sandler, Mark and Howard, Andrew and Zhu, Menglong and Zhmoginov, Andrey and Chen, Liang-Chieh},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={4510--4520},
  year={2018}
}

@TechReport{zb2017hand,
  author={Christian Zimmermann and Thomas Brox},
  title={Learning to Estimate 3D Hand Pose from Single RGB Images},
  institution={arXiv:1705.01389},
  year={2017},
  note="https://arxiv.org/abs/1705.01389",
  url="https://lmb.informatik.uni-freiburg.de/projects/hand3d/"
}

Results on RHD test set

Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
pose_mobilenet_v2	256x256	0.985	0.883	2.80	ckpt	log

Topdown Heatmap + Resnet on Rhd2d¶

@inproceedings{xiao2018simple,
  title={Simple baselines for human pose estimation and tracking},
  author={Xiao, Bin and Wu, Haiping and Wei, Yichen},
  booktitle={Proceedings of the European conference on computer vision (ECCV)},
  pages={466--481},
  year={2018}
}

@inproceedings{he2016deep,
  title={Deep residual learning for image recognition},
  author={He, Kaiming and Zhang, Xiangyu and Ren, Shaoqing and Sun, Jian},
  booktitle={Proceedings of the IEEE conference on computer vision and pattern recognition},
  pages={770--778},
  year={2016}
}

@TechReport{zb2017hand,
  author={Christian Zimmermann and Thomas Brox},
  title={Learning to Estimate 3D Hand Pose from Single RGB Images},
  institution={arXiv:1705.01389},
  year={2017},
  note="https://arxiv.org/abs/1705.01389",
  url="https://lmb.informatik.uni-freiburg.de/projects/hand3d/"
}

Results on RHD test set

Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
pose_resnet50	256x256	0.991	0.898	2.33	ckpt	log

Topdown Heatmap + Hrnetv2 + Udp on Rhd2d¶

@article{WangSCJDZLMTWLX19,
  title={Deep High-Resolution Representation Learning for Visual Recognition},
  author={Jingdong Wang and Ke Sun and Tianheng Cheng and
          Borui Jiang and Chaorui Deng and Yang Zhao and Dong Liu and Yadong Mu and
          Mingkui Tan and Xinggang Wang and Wenyu Liu and Bin Xiao},
  journal={TPAMI},
  year={2019}
}

@InProceedings{Huang_2020_CVPR,
  author = {Huang, Junjie and Zhu, Zheng and Guo, Feng and Huang, Guan},
  title = {The Devil Is in the Details: Delving Into Unbiased Data Processing for Human Pose Estimation},
  booktitle = {The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
  month = {June},
  year = {2020}
}

@TechReport{zb2017hand,
  author={Christian Zimmermann and Thomas Brox},
  title={Learning to Estimate 3D Hand Pose from Single RGB Images},
  institution={arXiv:1705.01389},
  year={2017},
  note="https://arxiv.org/abs/1705.01389",
  url="https://lmb.informatik.uni-freiburg.de/projects/hand3d/"
}

Results on RHD test set

Arch	Input Size	PCKh@0.7	AUC	EPE	ckpt	log
pose_hrnetv2_w18_udp	256x256	0.992	0.902	2.21	ckpt	log

Topdown Heatmap + Hrnetv2 + Dark on Rhd2d¶

@article{WangSCJDZLMTWLX19,
  title={Deep High-Resolution Representation Learning for Visual Recognition},
  author={Jingdong Wang and Ke Sun and Tianheng Cheng and
          Borui Jiang and Chaorui Deng and Yang Zhao and Dong Liu and Yadong Mu and
          Mingkui Tan and Xinggang Wang and Wenyu Liu and Bin Xiao},
  journal={TPAMI},
  year={2019}
}

@inproceedings{zhang2020distribution,
  title={Distribution-aware coordinate representation for human pose estimation},
  author={Zhang, Feng and Zhu, Xiatian and Dai, Hanbin and Ye, Mao and Zhu, Ce},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  pages={7093--7102},
  year={2020}
}

@TechReport{zb2017hand,
  author={Christian Zimmermann and Thomas Brox},
  title={Learning to Estimate 3D Hand Pose from Single RGB Images},
  institution={arXiv:1705.01389},
  year={2017},
  note="https://arxiv.org/abs/1705.01389",
  url="https://lmb.informatik.uni-freiburg.de/projects/hand3d/"
}

Results on RHD test set

Arch	Input Size	PCK@0.2	AUC	EPE	ckpt	log
pose_hrnetv2_w18_dark	256x256	0.992	0.903	2.17	ckpt	log

Topdown Heatmap + Hrnetv2 on Rhd2d¶

@article{WangSCJDZLMTWLX19,
  title={Deep High-Resolution Representation Learning for Visual Recognition},
  author={Jingdong Wang and Ke Sun and Tianheng Cheng and
          Borui Jiang and Chaorui Deng and Yang Zhao and Dong Liu and Yadong Mu and
          Mingkui Tan and Xinggang Wang and Wenyu Liu and Bin Xiao},
  journal={TPAMI},
  year={2019}
}