2D Human Pose Estimation on MPII (val)
98HeadViTPose-G
Evaluation results
| Method | Links | |||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ViTPose-GBackbone=ViTAE-G, Resolution=576x432, Bbox Source=ground truth2022.04 | 98 | 97.6 | 94.5 | 91.9 | 92.9 | 93 | 90.2 | 94.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ViTPose-LBackbone=ViT-L, Resolution=256x192, Bbox Source=ground truth2022.04 | 97.8 | 97.6 | 94.3 | 91.2 | 93 | 92.5 | 89.8 | 94 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| HRNet-W32Testing Scale=Multi-scale testing2019.02 | 97.7 | 96.3 | 90.9 | 86.7 | 89.7 | 87.4 | 84.1 | 90.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ViTPose-HBackbone=ViT-H, Resolution=256x192, Bbox Source=ground truth2022.04 | 97.7 | 97.6 | 94.4 | 91.5 | 93.2 | 92.6 | 90.3 | 94.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SimpleBaselineTesting Scale=Multi-scale testing2019.02 | 97.5 | 96.1 | 90.5 | 85.4 | 90.1 | 85.7 | 82.3 | 90.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ViTPose-BBackbone=ViT-B, Resolution=256x192, Bbox Source=ground truth2022.04 | 97.5 | 97.4 | 93.7 | 90.5 | 92.3 | 91.5 | 88.1 | 93.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PCTBackbone=Swin-Base, Input size=256 x 2562023.03 | 97.5 | 97.2 | 92.8 | 88.4 | 92.4 | 89.6 | 87.1 | 92.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| HourglassNetworkIdentification=oracle2018.03 | 97.44 | 98.27 | 94.02 | 92.22 | 93.3 | 90.49 | 86.02 | 93.22 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Hierarchical, Parallel & Multi-Scale Residual Blockprecision=real, scale=8x HG, number of parameters=25M2017.03 | 97.4 | 96 | 90.7 | 86.2 | 89.6 | 86.1 | 83.2 | — | 89.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Hierarchical, Parallel & Multi-Scale Residual Block [8x] (real-valued)# parameters=25M, precision=real-valued, HG networks=82018.08 | 97.4 | 96 | 90.7 | 86.2 | 89.6 | 86.1 | 83.2 | — | 89.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Yang et al.Testing Scale=Multi-scale testing2019.02 | 97.4 | 96.2 | 91.1 | 86.9 | 90.1 | 86 | 83.9 | 90.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Tang et al.Testing Scale=Multi-scale testing2019.02 | 97.4 | 96.2 | 91 | 86.9 | 90.6 | 86.8 | 84.5 | 90.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| OmniPoseModule=WASPv2, Params (M)=68.1, GFLOPS=22.62021.03 | 97.4 | 97.1 | 92.4 | 88.7 | 91.2 | 89.9 | 85.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 92.3 | — | |
| OmniPoseModule=WASP, Params (M)=68.2, GFLOPS=23.02021.03 | 97.4 | 96.6 | 91.9 | 87.2 | 90.1 | 88 | 83.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 91.2 | — | |
| Stacked Hourglass [22]number of parameters=25M2017.03 | 97.3 | 96 | 90.2 | 85.2 | 89.1 | 85.1 | 82 | — | 89.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Stacked Hourglass [2]# parameters=25M2018.08 | 97.3 | 96 | 90.2 | 85.2 | 89.1 | 85.1 | 82 | — | 89.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ScaleNASBackbone=ScaleNet-P1, #Params=28.5M, GFLOPS=9.3, Input size=256x2562020.11 | 97.3 | 96.5 | 91.5 | 87.3 | 90 | 87.5 | — | 91 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PRTRBackbone=HRNet-W32, Prediction Type=Regression Based2021.04 | 97.3 | 96 | 90.6 | 84.5 | 89.7 | 85.5 | 79 | 89.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ASDABackbone=HRNet-w48, Resolution=256x256, Bbox Source=ground truth2022.04 | 97.3 | 96.5 | 91.7 | 87.9 | 90.8 | 88.2 | 84.2 | 91.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PRTRInput size=256 x 2562023.03 | 97.3 | 96 | 90.6 | 84.5 | 89.7 | 85.5 | 79 | 89.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DarkPoseParams (M)=63.6, GFLOPS=19.52021.03 | 97.2 | 95.9 | 91.2 | 86.7 | 89.7 | 86.7 | 84 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 90.6 | — | |
| DARKInput size=256 x 2562023.03 | 97.2 | 95.9 | 91.2 | 86.7 | 89.7 | 86.7 | 84 | 90.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SimCCInput size=256 x 2562023.03 | 97.2 | 96 | 90.4 | 85.6 | 89.5 | 85.8 | 81.8 | 90 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| HRNet-W32Testing Scale=Single-scale testing2019.02 | 97.1 | 95.9 | 90.3 | 86.4 | 89.1 | 87.1 | 83.3 | 90.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Newell et al.Testing Scale=Multi-scale testing2019.02 | 97.1 | 96.1 | 90.8 | 86.2 | 89.9 | 85.9 | 83.5 | 90 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| HRNetBackbone=HRNet-W32, #Params=28.5M, GFLOPS=9.5, Input size=256x2562020.11 | 97.1 | 95.9 | 90.3 | 86.4 | 89.1 | 87.1 | — | 90.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| HRNetParams (M)=63.6, GFLOPS=19.52021.03 | 97.1 | 95.9 | 90.3 | 86.5 | 89.1 | 87.1 | 83.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 90.3 | — | |
| HRNetBackbone=HRNet-W32, Prediction Type=Heatmap Based2021.04 | 97.1 | 95.9 | 90.3 | 86.4 | 89.1 | 87.1 | 83.3 | 90.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DPIT-B#Params=21.6M, Input size=256 x 2562022.09 | 97.1 | 95.7 | 90 | 84.6 | 89.4 | 85.9 | 80.7 | 89.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| HRNetBackbone=HRNet-w48, Resolution=256x256, Bbox Source=ground truth2022.04 | 97.1 | 95.8 | 90.7 | 85.6 | 89 | 86.8 | 82.1 | 90.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| HRNetInput size=256 x 2562023.03 | 97.1 | 95.9 | 90.3 | 86.4 | 89.1 | 87.1 | 83.3 | 90.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| TokenPoseInput size=256 x 2562023.03 | 97.1 | 95.9 | 90.4 | 86 | 89.3 | 87.1 | 82.5 | 90.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SimpleBaselineTesting Scale=Single-scale testing2019.02 | 97 | 95.9 | 90.3 | 85 | 89.2 | 85.3 | 81.3 | 89.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Simple BaselineBackbone=ResNet-152, Prediction Type=Heatmap Based2021.04 | 97 | 95.9 | 90.3 | 85 | 89.2 | 85.3 | 81.3 | 89.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SimpleBaselineBackbone=R-152, #Params=68.6M, Input size=256 x 2562022.09 | 97 | 95.9 | 90 | 85 | 89.2 | 85.3 | 81.3 | 89.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SimBaInput size=256 x 2562023.03 | 97 | 95.6 | 90 | 86.2 | 89.7 | 86.9 | 82.9 | 90.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SimpleBaselineBackbone=R-101, #Params=53.0M, Input size=256 x 2562022.09 | 96.9 | 95.9 | 89.5 | 84.4 | 88.4 | 84.5 | 80.7 | 89.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| HRNetBackbone=H-W32, #Params=28.5M, Input size=256 x 2562022.09 | 96.9 | 96 | 90.6 | 85.8 | 88.7 | 86.6 | 82.6 | 90.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| HRNetBackbone=HRNet-w32, Resolution=256x256, Bbox Source=ground truth2022.04 | 96.9 | 95.9 | 90.5 | 85.9 | 89.1 | 86.1 | 82.5 | 90 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Hierarchical, Parallel & Multi-Scale Residual Blockprecision=real, scale=1x HG, number of parameters=6M2017.03 | 96.8 | 93.8 | 86.4 | 80.3 | 87 | 80.4 | 75.7 | — | 85.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Hierarchical, Parallel & Multi-Scale Residual Block [1x] (real-valued)# parameters=6M, precision=real-valued, HG networks=12018.08 | 96.8 | 93.8 | 86.4 | 80.3 | 87 | 80.4 | 75.7 | — | 85.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Yang et al.Testing Scale=Single-scale testing2019.02 | 96.8 | 96 | 90.4 | 86 | 89.5 | 85.2 | 82.3 | 89.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DPIT-L/D6#Params=31.8M, Input size=256 x 2562022.09 | 96.7 | 95.9 | 90.8 | 85.9 | 89.2 | 86 | 82.6 | 90.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| OmniPose-LiteParams (M)=19.4, GFLOPS=5.82021.03 | 96.6 | 95.8 | 89.1 | 84.3 | 89 | 84.1 | 79.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 89 | — | |
| Newell et al.Testing Scale=Single-scale testing2019.02 | 96.5 | 96 | 90.3 | 85.4 | 88.8 | 85 | 81.9 | 89.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| HourglassNetworkIdentification=default2018.03 | 96.49 | 95.38 | 89.16 | 84.89 | 87.73 | 84.08 | 80.3 | 88.39 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SimpleBaselineBackbone=ResNet-152, #Params=68.6M, GFLOPS=20.9, Input size=256x2562020.11 | 96.4 | 95.3 | 89 | 83.2 | 88.4 | 84 | — | 88.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PRTRBackbone=ResNet-152, Prediction Type=Regression Based2021.04 | 96.4 | 94.9 | 88.4 | 82.6 | 88.6 | 84.1 | 78.4 | 88.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SimpleBaselineBackbone=R-50, #Params=34.0M, Input size=256 x 2562022.09 | 96.4 | 95.3 | 89 | 83.2 | 88.4 | 84 | 79.6 | 88.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PRTRBackbone=ResNet-101, Prediction Type=Regression Based2021.04 | 96.3 | 95 | 88.3 | 82.4 | 88.1 | 83.6 | 77.4 | 87.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Convolutional Pose MachinesBackbone=CPM, Prediction Type=Heatmap Based2021.04 | 96.2 | 95 | 87.5 | 82.2 | 87.6 | 82.7 | 78.4 | 87.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CFABackbone=ResNet-101, Resolution=384x384, Bbox Source=ground truth2022.04 | 95.9 | 95.4 | 91 | 86.9 | 89.8 | 87.6 | 83.9 | 90.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Tang et al.Testing Scale=Single-scale testing2019.02 | 95.6 | 95.9 | 90.7 | 86.5 | 89.9 | 86.6 | 82.5 | 89.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Hierarchical, Parallel & Multi-Scale Residual Blockprecision=binary, number of parameters=6M2017.03 | 94.7 | 89.6 | 78.8 | 71.5 | 79.1 | 70.5 | 64 | — | 78.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Hierarchical, Parallel & Multi-Scale Residual Block (binarized)# parameters=6M, precision=binary2018.08 | 94.7 | 89.6 | 78.8 | 71.5 | 79.1 | 70.5 | 64 | — | 78.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CMU Pose2021.03 | 92.4 | 90.4 | 80.9 | 70.8 | 79.5 | 73.1 | 66.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 79.1 | — | |
| StarMapIdentification=oracle2018.03 | 92.12 | 93.65 | 90.49 | 86.09 | 82.4 | 87.23 | 82.22 | 88.17 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SPM2021.03 | 92 | 88.5 | 78.6 | 69.4 | 77.7 | 73.8 | 63.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 77.7 | — | |
| StarMapIdentification=learned2018.03 | 91 | 88.69 | 83.02 | 73.58 | 74.16 | 76.67 | 69.01 | 79.85 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RMPE2021.03 | 88.4 | 86.5 | 78.6 | 70.4 | 74.4 | 73 | 65.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 76.7 | — | |
| SimpleBaselineBackbone=ResNet-152, Resolution=256x256, Bbox Source=ground truth2022.04 | 86.9 | 95.4 | 89.4 | 84 | 88 | 84.6 | 82.1 | 89 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DARKBackbone=HRNet-W32, Input size=256x256, Evaluation Protocol=Single-scale2019.10 | — | — | — | — | — | — | — | — | — | 97.2 | 95.9 | 91.2 | 86.7 | 89.7 | 86.7 | 84 | 90.6 | 55.2 | 47.8 | 47.4 | 45.2 | 20.1 | 33.4 | 35.4 | 42 | — | — | |
| Dite-HRNetBackbone=Dite-HRNet-30, TTA=Yes, #Params=1.8M, FLOPs=0.40G, Input Size=256 x 2562025.12 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 87.6 | |
| EL-HRNetBackbone=EL-HRNet-w32, TTA=No, #Params=5.0M, FLOPs=2.66G, Input Size=256 x 2562025.12 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 87.7 | |
| HBC#parameters=6.2M2019.04 | — | — | — | — | — | — | — | — | 78.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| HF-HRNetBackbone=HF-HRNet-18, TTA=Yes, #Params=4.6M, FLOPs=0.90G, Input Size=256 x 2562025.12 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 88 | |
| HF-HRNetBackbone=HF-HRNet-30, TTA=Yes, #Params=7.4M, FLOPs=1.50G, Input Size=256 x 2562025.12 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 88.5 | |
| HRN32Backbone=HRNet-W32, Input size=256x256, Evaluation Protocol=Single-scale2019.10 | — | — | — | — | — | — | — | — | — | 97.1 | 95.9 | 90.3 | 86.5 | 89.1 | 87.1 | 83.3 | 90.3 | 51.1 | 42.7 | 42 | 41.6 | 17.9 | 29.9 | 31 | 37.7 | — | — | |
| HRNetBackbone=HRNet-W32, Approach Type=HM2022.01 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 90.1 | — | — | — | — | — | — | — | — | — | — | |
| HRNetBackbone=HRNet-w32, Input Size=256 × 256, GFLOPs=10.27, Flip test=true2023.03 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 90 | |
| HRNetBackbone=HRNet-W48, Datasets=1, Input size=256x2562025.05 | — | — | — | — | — | — | — | — | 90.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| IntegralBackbone=ResNet-101, Prediction Type=Regression Based2021.04 | — | — | — | — | — | — | — | 87.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| IntegralBackbone=ResNet-101, Approach Type=Reg2022.01 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 87.3 | — | — | — | — | — | — | — | — | — | — | |
| LAPBackbone=Hourglass, TTA=Yes, #Params=2.34M, FLOPs=3.70G, Input Size=256 x 2562025.12 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 87.6 | |
| LAPXBackbone=Hourglass, TTA=No, #Params=2.30M, FLOPs=3.45G, Input Size=256 x 2562025.12 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 88 | |
| LAPXBackbone=Hourglass, TTA=Yes, #Params=2.30M, FLOPs=3.45G, Input Size=256 x 2562025.12 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 88.6 | |
| Lite-HRNetBackbone=Lite-HRNet-30, TTA=Yes, #Params=1.8M, FLOPs=0.42G, Input Size=256 x 2562025.12 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 87 | |
| Lite-HRNet-18input size=256 x 256, #Params=1.1M, GFLOPS=0.272021.04 | — | — | — | — | — | — | — | — | 86.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Lite-HRNet-30input size=256 x 256, #Params=1.8M, GFLOPS=0.422021.04 | — | — | — | — | — | — | — | — | 87 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LMFormerBackbone=LMFormer-L, TTA=No, #Params=4.1M, FLOPs=1.90G, Input Size=256 x 2562025.12 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 87.6 | |
| MobileNet V2 1xinput size=256 x 256, #Params=9.6M, GFLOPS=1.972021.04 | — | — | — | — | — | — | — | — | 85.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MobileNetV3 1xinput size=256 x 256, #Params=8.7M, GFLOPS=1.822021.04 | — | — | — | — | — | — | — | — | 84.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Ours#parameters=6.0M2019.04 | — | — | — | — | — | — | — | — | 82.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PCTBackbone=Swin-B, Datasets=1, Input size=256x2562025.05 | — | — | — | — | — | — | — | — | 92.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PoseBHBackbone=ViT-B, Datasets=62025.05 | — | — | — | — | — | — | — | — | 93.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PoseBHBackbone=ViT-H, Datasets=62025.05 | — | — | — | — | — | — | — | — | 94.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PoseurBackbone=HRNet-W32, Approach Type=Reg2022.01 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 90.5 | — | — | — | — | — | — | — | — | — | — | |
| PRTRBackbone=HRNet-W32, Approach Type=Reg2022.01 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 89.5 | — | — | — | — | — | — | — | — | — | — | |
| Real valued#parameters=6.0M2019.04 | — | — | — | — | — | — | — | — | 85.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RTMPose-mBackbone=CSPNeXt-m, Input Size=256 × 256, GFLOPs=2.57, Flip test=true, Pre-trained=false2023.03 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 88.9 | |
| RTMPose-mBackbone=CSPNeXt-m, Input Size=256 × 256, GFLOPs=2.57, Flip test=true, Pre-trained=AIC+COCO, Fine-tuned=MPII2023.03 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 90.7 | |
| ShuffleNetV2 1xinput size=256 x 256, #Params=7.6M, GFLOPS=1.702021.04 | — | — | — | — | — | — | — | — | 82.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SimBaBackbone=ResNet-152, Approach Type=HM2022.01 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 89.6 | — | — | — | — | — | — | — | — | — | — | |
| SimCCBackbone=HRNet-w32, Input Size=256 × 256, GFLOPs=10.34, Flip test=true2023.03 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 90 | |
| SimCCBackbone=HRNet-W48, Datasets=12025.05 | — | — | — | — | — | — | — | — | 90 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SimpleBaselineBackbone=ResNet-50, Input Size=256 × 256, GFLOPs=7.28, Flip test=true2023.03 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 88.2 | |
| Small HRNet-W16input size=256 x 256, #Params=1.3M, GFLOPS=0.722021.04 | — | — | — | — | — | — | — | — | 80.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| TokenPoseBackbone=L/D24, Approach Type=HM2022.01 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 90.2 | — | — | — | — | — | — | — | — | — | — | |
| TokenPoseBackbone=L/D24, Input Size=256 × 256, GFLOPs=11, Flip test=true2023.03 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 90.2 | |
| TransPose-H-A6Backbone=HRNet-w48, Resolution=256x256, Bbox Source=ground truth2022.04 | — | — | — | — | — | — | — | 92.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — |