Whole-body Pose Estimation on COCO-Wholebody 1.0 (val)
75.9Body APViTPose++-H
Evaluation results
| Method | Links | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ViTPose++-HBackbone=ViT-H, Resolution=256x1922022.12 | 75.9 | — | 77.9 | — | 63.3 | — | 54.7 | — | 61.2 | — | — | — | — | |
| ViTPose++-LBackbone=ViT-L, Resolution=256x1922022.12 | 75.3 | — | 77.1 | — | 63 | — | 54.2 | — | 60.6 | — | — | — | — | |
| ZoomNetCategory=Top-down, Input Size=384x288, GFLOPs=28.52023.03 | 74.5 | 81 | 60.9 | 70.8 | 88 | 92.4 | 57.9 | 73.4 | 63 | 74.2 | — | — | — | |
| ZoomNetMethod paradigm=Top-down2021.10 | 74.3 | — | 79.8 | — | 62.3 | — | 40.1 | — | 54.1 | — | — | — | — | |
| ZoomNetResolution=384 x 288, training set=COCO-WholeBody V0.52022.04 | 74.3 | 80.2 | 79.8 | 86.9 | 62.3 | 70.1 | 40.1 | 49.8 | 54.1 | 65.8 | — | — | — | |
| ZoomNetBackbone=HRNet, Resolution=384x2882022.12 | 74.3 | — | 79.8 | — | 62.3 | — | 40.1 | — | 54.1 | — | — | — | — | |
| ZoomNetCategory=Top-down methods, Time complexity note=run time linearly increases as the number of people in an image increases2021.06 | 74.3 | 80.2 | 79.8 | 86.9 | 62.3 | 70.1 | 40.1 | 49.8 | 54.1 | 65.8 | 62.1 | 70.6 | 175 | |
| ZoomNASCategory=Top-down, Input Size=384x288, GFLOPs=18.02023.03 | 74 | 80.7 | 61.7 | 71.8 | 88.9 | 93 | 62.5 | 74 | 65.4 | 74.4 | — | — | — | |
| ViTPose-HBackbone=ViT-H, Resolution=256x1922022.12 | 73.5 | — | 76.4 | — | 63.2 | — | 54.9 | — | 60 | — | — | — | — | |
| ViTPose++-BBackbone=ViT-B, Resolution=256x1922022.12 | 73.3 | — | 74.2 | — | 60.1 | — | 50.2 | — | 57.4 | — | — | — | — | |
| TCFormer-LargeResolution=384 x 2882022.04 | 73.1 | 80.3 | 75.2 | 85.5 | 77.4 | 84.5 | 60.7 | 71.2 | 64.4 | 73.5 | — | — | — | |
| ViTPose-LBackbone=ViT-L, Resolution=256x1922022.12 | 72.9 | — | 75.2 | — | 63.9 | — | 54.5 | — | 59.6 | — | — | — | — | |
| HRNet-w48Resolution=384 x 2882022.04 | 72.2 | 79 | 69.4 | 79.9 | 77.7 | 83.4 | 58.7 | 67.9 | 63.1 | 71.6 | — | — | — | |
| ViTPose++-SBackbone=ViT-S, Resolution=256x1922022.12 | 71.6 | — | 72.1 | — | 55.9 | — | 45.3 | — | 54.4 | — | — | — | — | |
| RTMPose-xCategory=Top-down, Input Size=384x288, GFLOPs=18.1, Pre-trained on AIC+COCO=true2023.03 | 71.4 | 78.4 | 69.2 | 81 | 88.9 | 92.3 | 59 | 68.5 | 65.3 | 73.3 | — | — | — | |
| RTMPose-lCategory=Top-down, Input Size=384x288, GFLOPs=10.1, Pre-trained on AIC+COCO=true2023.03 | 71.2 | 78.1 | 69.3 | 81.1 | 88.2 | 91.9 | 57.9 | 67.7 | 64.8 | 73 | — | — | — | |
| RTMPose-xCategory=Top-down, Input Size=384x288, GFLOPs=18.1, Pre-trained on AIC+COCO=false2023.03 | 71.2 | 78 | 68.1 | 80.4 | 89 | 92.2 | 59.3 | 68.7 | 65.2 | 73.2 | — | — | — | |
| FastPose50-dcn-siCategory=Top-down, Input Size=256x192, GFLOPs=6.12023.03 | 70.6 | 75.6 | 70.2 | 77.5 | 77.5 | 82.5 | 45.7 | 53.9 | 59.2 | 66.5 | — | — | — | |
| SBL-Res152Resolution=384 x 2882022.04 | 70.3 | 78 | 69.3 | 81.3 | 75.1 | 82.5 | 55.9 | 66.7 | 61 | 70.5 | — | — | — | |
| Hulk-FTBackbone=ViT-B, Input Resolution=256x1922023.12 | 70.2 | 77.5 | 65.9 | 79.1 | 84.2 | 90.1 | 53.2 | 63.1 | 61.6 | 72.1 | — | — | — | |
| HRNetBackbone=HRNet-W48, Input Resolution=256x1922023.12 | 70.1 | 77.6 | 67.5 | 78.7 | 65.6 | 74.3 | 53.5 | 63.9 | 57.9 | 68.1 | — | — | — | |
| HRNetCategory=Top-down, Input Size=384x288, GFLOPs=16.02023.03 | 70.1 | 77.3 | 58.6 | 69.2 | 72.7 | 78.3 | 51.6 | 60.4 | 58.6 | 67.4 | — | — | — | |
| HRNetBackbone=HRNet-w32, Resolution=256x1922022.12 | 70 | — | 56.7 | — | 63.7 | — | 47.3 | — | 55.3 | — | — | — | — | |
| HRNetBackbone=HRNet-w48, Resolution=256x1922022.12 | 70 | — | 67.2 | — | 65.6 | — | 53.4 | — | 57.9 | — | — | — | — | |
| Poseur + CCNetBase Model=Poseur, Module=CCNet, Confidence Strategy=Aggregation of predicted part2023.11 | 69.9 | 72.3 | 69.2 | 72.9 | 86.4 | 88.1 | 62.7 | 65.4 | 63.3 | 67.2 | — | — | — | |
| ViTPose-BBackbone=ViT-B, Input Resolution=256x1922023.12 | 69.6 | — | 70.1 | — | 62.1 | — | 50.8 | — | 56.3 | — | — | — | — | |
| Keypoint CommunitiesMethod paradigm=Bottom-up, Backbone=ShuffleNetV22021.10 | 69.6 | — | 63.4 | — | 85 | — | 52.9 | — | 60.4 | — | — | — | — | |
| ViTPose-BBackbone=ViT-B, Resolution=256x1922022.12 | 69.6 | — | 70.1 | — | 62.1 | — | 50.8 | — | 56.3 | — | — | — | — | |
| RTMPose-lCategory=Top-down, Input Size=256x192, GFLOPs=4.5, Pre-trained on AIC+COCO=true2023.03 | 69.5 | 76.9 | 65.8 | 78.5 | 83.3 | 88.7 | 51.9 | 62.8 | 61.1 | 70 | — | — | — | |
| FastPoseBackbone=ResNet50-dcn, Resolution=256x1922022.12 | 69.3 | — | 69 | — | 75.9 | — | 45.3 | — | 57.7 | — | — | — | — | |
| TCFormerBackbone=TCFormer, Input Resolution=256x1922023.12 | 69.1 | 77 | 69.8 | 81.3 | 64.9 | 74.6 | 53.5 | 65 | 57.2 | 67.8 | — | — | — | |
| DWPose-mBackbone=CSPNeXt-m, Input Resolution=256x1922023.12 | 68.5 | 76.1 | 63.6 | 77.2 | 82.8 | 88.1 | 52.7 | 63.4 | 60.6 | 69.5 | — | — | — | |
| PoseurBase Model=Poseur, Confidence Strategy=Part-wise confidence2023.11 | 68.5 | — | 68.9 | — | 85.9 | — | 62.5 | — | 61 | — | — | — | — | |
| SimpleBaseBackbone=ResNet-152, Input Resolution=256x1922023.12 | 68.2 | 76.4 | 66.1 | 78.7 | 62.3 | 72.8 | 48.1 | 60.7 | 54.8 | 66.1 | — | — | — | |
| HRNetBackbone=HRNet-W32, Input Resolution=256x1922023.12 | 67.8 | 75.5 | 54.3 | 66.1 | 63 | 70.8 | 46.7 | 56.6 | 53.6 | 63.6 | — | — | — | |
| RTMPose-mBackbone=CSPNeXt-m, Input Resolution=256x1922023.12 | 67.3 | 75 | 61.5 | 75.2 | 81.3 | 87.1 | 47.5 | 58.9 | 58.2 | 67.4 | — | — | — | |
| PVTCategory=Top-down, Input Size=384x288, GFLOPs=19.72023.03 | 67.3 | 76.1 | 66 | 79.4 | 74.5 | 82.2 | 54.5 | 65.4 | 58.9 | 68.9 | — | — | — | |
| RTMPose-mCategory=Top-down, Input Size=256x192, GFLOPs=2.2, Pre-trained on AIC+COCO=true2023.03 | 67.3 | 75 | 61.5 | 75.2 | 81.3 | 87.1 | 47.5 | 58.9 | 58.2 | 67.4 | — | — | — | |
| PoseurBase Model=Poseur, Confidence Strategy=Whole-instance confidence2023.11 | 67.2 | — | 63.6 | — | 84.6 | — | 58.3 | — | 61 | — | — | — | — | |
| SimpleBaselineBackbone=ResNet-101, Resolution=256x1922022.12 | 67 | — | 64 | — | 61.1 | — | 46.3 | — | 53.3 | — | — | — | — | |
| SimpleBaselineCategory=Top-down, Input Size=384x288, GFLOPs=20.42023.03 | 66.6 | 74.7 | 63.5 | 76.3 | 73.2 | 81.2 | 53.7 | 64.7 | 57.3 | 67.1 | — | — | — | |
| ViTPose-SBackbone=ViT-S, Resolution=256x1922022.12 | 66.1 | — | 64.5 | — | 59.8 | — | 45.9 | — | 52.3 | — | — | — | — | |
| HRNetMethod paradigm=Top-down2021.10 | 65.9 | — | 31.4 | — | 52.3 | — | 30 | — | 43.2 | — | — | — | — | |
| HRNetCategory=Top-down methods, Time complexity note=run time linearly increases as the number of people in an image increases2021.06 | 65.9 | 70.9 | 31.4 | 42.4 | 52.3 | 58.2 | 30 | 36.3 | 43.2 | 52 | 44.6 | 52 | — | |
| SimpleBaseBackbone=ResNet-50, Input Resolution=256x1922023.12 | 65.2 | 73.8 | 61.5 | 74.9 | 60.6 | 71.5 | 46 | 58.4 | 52.1 | 63.3 | — | — | — | |
| SimpleBaselineBackbone=ResNet-50, Resolution=256x1922022.12 | 65.2 | — | 61.4 | — | 60.8 | — | 46 | — | 52 | — | — | — | — | |
| HigherHRNet-w48Resolution=512 x 5122022.04 | 63 | 70.6 | 44 | 57.3 | 73 | 77.7 | 38.9 | 47.7 | 48.7 | 57.4 | — | — | — | |
| HPRNetCategory=Bottom-up methods, Backbone=HG (Hourglass-104)2021.06 | 59.4 | 68.3 | 53 | 65.4 | 75.4 | 86.8 | 50.4 | 64.2 | 34.8 | 49.2 | 54.6 | 66.8 | 101 | |
| AE+HRNet-w48Resolution=512 x 5122022.04 | 59.2 | 68.6 | 44.3 | 59.5 | 61.9 | 67.4 | 34.7 | 43.8 | 42.2 | 53.2 | — | — | — | |
| AECategory=Bottom-up, Input Size=512x512, GFLOPs=212.42023.03 | 58 | 66.1 | 57.7 | 72.5 | 58.8 | 65.4 | 48.1 | 57.4 | 44 | 54.5 | — | — | — | |
| OpenPoseMethod paradigm=Bottom-up2021.10 | 56.3 | — | 53.2 | — | 48.2 | — | 19.8 | — | 33.8 | — | — | — | — | |
| OpenPoseResolution=480 x 4802022.04 | 56.3 | 61.2 | 53.2 | 64.5 | 76.5 | 84 | 38.6 | 43.3 | 44.2 | 52.3 | — | — | — | |
| OpenPoseCategory=Whole-body, GFLOPs=451.12023.03 | 56.3 | 61.2 | 53.2 | 64.5 | 76.5 | 84 | 38.6 | 43.3 | 44.2 | 52.3 | — | — | — | |
| OpenPoseCategory=Top-down methods2021.06 | 56.3 | 61.2 | 53.2 | 64.5 | 48.2 | 62.6 | 19.8 | 34.2 | 33.8 | 44.9 | 42.3 | 53.5 | 45 | |
| HPRNetCategory=Bottom-up methods, Backbone=DLA2021.06 | 55.2 | 63.1 | 49.1 | 60.9 | 74.6 | 83.7 | 47 | 60.8 | 31.5 | 44.6 | 51.5 | 62.6 | 37 | |
| DeepPoseCategory=Top-down, Input Size=384x288, GFLOPs=17.32023.03 | 44.4 | 56.8 | 36.8 | 53.7 | 49.3 | 66.3 | 23.5 | 41 | 33.5 | 48.4 | — | — | — | |
| SNResolution=480 x 480, multi-scale testing=true2022.04 | 42.7 | 58.3 | 9.9 | 36.9 | 64.9 | 69.7 | 40.8 | 58 | 32.7 | 45.6 | — | — | — | |
| SNCategory=Whole-body, GFLOPs=272.3, multi-scale testing=true2023.03 | 42.7 | 58.3 | 9.9 | 36.9 | 64.9 | 69.7 | 40.8 | 58 | 32.7 | 45.6 | — | — | — | |
| AEMethod paradigm=Bottom-up2021.10 | 40.5 | — | 7.7 | — | 47.7 | — | 34.1 | — | 27.4 | — | — | — | — | |
| AECategory=Bottom-up methods2021.06 | 40.5 | 46.4 | 7.7 | 16 | 47.7 | 58 | 34.1 | 43.5 | 27.4 | 35 | 31.5 | 39.8 | — | |
| PAFResolution=480 x 480, multi-scale testing=true2022.04 | 38.1 | 52.6 | 5.3 | 27.8 | 65.5 | 70.1 | 35.9 | 52.8 | 29.5 | 40.5 | — | — | — | |
| PAFCategory=Bottom-up, Input Size=512x512, GFLOPs=329.1, multi-scale testing=true2023.03 | 38.1 | 52.6 | 5.3 | 27.8 | 65.6 | 70.1 | 35.9 | 52.8 | 29.5 | 40.5 | — | — | — | |
| SNCategory=Bottom-up methods2021.06 | 28 | 33.6 | 12.1 | 27.7 | 38.2 | 44 | 13.8 | 33.6 | 16.1 | 20.9 | 21.6 | 32 | 216 | |
| PAFCategory=Bottom-up methods2021.06 | 26.6 | 32.8 | 10 | 25.7 | 30.9 | 36.2 | 13.3 | 32.1 | 14.1 | 18.5 | 19 | 29.1 | 100 | |
| HRNet-w32+DARKInput Size=256 x 192, GFLOPS=7.72, CPU(ms)=39.051, GPU(ms)=5.154, Flip test=not used2023.03 | — | — | — | — | — | — | — | — | — | — | 57.8 | — | — | |
| HRNet-w48+DARKInput Size=384 x 288, GFLOPS=35.52, CPU(ms)=150.765, GPU(ms)=13.974, Flip test=not used2023.03 | — | — | — | — | — | — | — | — | — | — | 65.3 | — | — | |
| PoseBHBackbone=ViT-Base, Input Size=256x1922025.05 | — | — | — | — | — | — | — | — | — | — | 57.9 | — | — | |
| RTMPose-lInput Size=256 x 192, GFLOPS=4.52, CPU(ms)=23.41, GPU(ms)=5.673, Flip test=not used2023.03 | — | — | — | — | — | — | — | — | — | — | 62.2 | — | — | |
| RTMPose-lInput Size=384 x 288, GFLOPS=10.07, CPU(ms)=44.581, GPU(ms)=7.678, Flip test=not used2023.03 | — | — | — | — | — | — | — | — | — | — | 66.1 | — | — | |
| RTMPose-mInput Size=256 x 192, GFLOPS=2.22, CPU(ms)=13.496, GPU(ms)=4, Flip test=not used2023.03 | — | — | — | — | — | — | — | — | — | — | 59.1 | — | — | |
| UniHCPBackbone=ViT-Base, Input Size=256x1922025.05 | — | — | — | — | — | — | — | — | — | — | 20.1 | — | — | |
| ViTPose++Backbone=ViT-Base, Input Size=256x1922025.05 | — | — | — | — | — | — | — | — | — | — | 57.1 | — | — |