Human Pose Estimation on PoseTrack 2018 (val)
89.8Total ScoreTAR-ViTPose
Evaluation results
| Method | Links | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| TAR-ViTPoseBackbone=ViT-H, Bounding box source=ground-truth ones, Temporal span T=22026.03 | 89.8 | 90.9 | 93.1 | 91.1 | 88.5 | 84.7 | 90.4 | 89 | — | — | — | — | — | — | |
| MTPoseBackbone=ViT (not specified), Bounding box source=ground-truth ones2026.03 | 89 | 89.4 | 92.4 | 90.1 | 87.3 | 85.7 | 89.7 | 88.1 | — | — | — | — | — | — | |
| TAR-ViTPoseBackbone=ViT-B, Bounding box source=ground-truth ones, Temporal span T=22026.03 | 88 | 88.7 | 91.8 | 89.4 | 86.1 | 83.6 | 88.5 | 87.3 | — | — | — | — | — | — | |
| PoseidonImage Size=288x384, Temporal window=2 previous and 2 next frames2025.01 | 87.8 | 88.8 | 91.4 | 88.6 | 86.3 | 83.3 | 88.8 | 87.2 | — | — | — | — | — | — | |
| PoseidonBackbone=ViT-H, Bounding box source=ground-truth ones2026.03 | 87.8 | 88.8 | 91.4 | 88.6 | 86.3 | 83.3 | 88.8 | 87.2 | — | — | — | — | — | — | |
| GLSMambaBackbone=ViT-H, Bounding box source=unspecified sources or ground-truth ones2026.03 | 84.9 | 85.6 | 88.9 | 86.5 | 83.6 | 82.9 | 85.7 | 81.4 | — | — | — | — | — | — | |
| CM-PoseBackbone=ViT (not specified), Bounding box source=unspecified sources or ground-truth ones2026.03 | 84.4 | 85.7 | 88.9 | 85.8 | 81 | 84.4 | 84.2 | 80.1 | — | — | — | — | — | — | |
| TAR-ViTPoseBackbone=ViT-H, Bounding box source=Faster R-CNN detector [28], Temporal span T=22026.03 | 84.2 | 86.8 | 88.9 | 86.1 | 81.9 | 82.9 | 83.4 | 77.9 | — | — | — | — | — | — | |
| JM-Pose2025.01 | 84.1 | 86.6 | 88.7 | 86 | 81.6 | 83.3 | 83.2 | 78.2 | — | — | — | — | — | — | |
| TDMI-ST2023.03 | 83.6 | 86.7 | 88.9 | 85.4 | 80.6 | 82.4 | 82.1 | 77.6 | — | — | — | — | — | — | |
| TDMI-ST2025.01 | 83.6 | 86.7 | 88.9 | 85.4 | 80.6 | 82.4 | 82.1 | 77.6 | — | — | — | — | — | — | |
| TDMI2023.03 | 83.5 | 86.2 | 88.7 | 85.4 | 80.6 | 82.4 | 82.1 | 77.5 | — | — | — | — | — | — | |
| TDMIBackbone=HRNet-W48, Bounding box source=unspecified sources or ground-truth ones2026.03 | 83.5 | 86.2 | 88.7 | 85.4 | 80.6 | 82.4 | 82.1 | 77.5 | — | — | — | — | — | — | |
| DSTABackbone=ViT-H, Method category=regression-based, Temporal span T=22024.03 | 83.4 | 85.9 | 88.8 | 85 | 81.1 | 81.5 | 83 | 77.4 | — | — | — | — | — | — | |
| DSTA2025.01 | 83.4 | 85.9 | 88.8 | 85 | 81.1 | 81.5 | 83 | 77.4 | — | — | — | — | — | — | |
| DSTABackbone=ViT-H, Bounding box source=Faster R-CNN detector [28], Temporal span T=22026.03 | 83.4 | 85.9 | 88.8 | 85 | 81.1 | 81.5 | 83 | 77.4 | — | — | — | — | — | — | |
| S-VIPNASBackbone=ResNet-50, Image Size=256 x 192, #Params=7.3M, GFLOPS=1.442021.05 | 83.2 | 88.1 | 89.6 | 84.5 | 77.4 | 81.1 | 81.8 | 77.6 | — | — | — | — | — | — | |
| LightTrackBackbone=ResNet152, Image Size=384 x 288, #Params=68.6M, GFLOPS=35.62021.05 | 83 | 89.4 | 88.5 | 84.4 | 76.2 | 81.2 | 80.5 | 77.5 | — | — | — | — | — | — | |
| DiffPose2025.01 | 83 | 85 | 87.7 | 84.3 | 81.5 | 81.4 | 82.9 | 77.6 | — | — | — | — | — | — | |
| DiffPoseBackbone=ViT (not specified), Bounding box source=unspecified sources or ground-truth ones2026.03 | 83 | 85 | 87.7 | 84.3 | 81.5 | 81.4 | 82.9 | 77.6 | — | — | — | — | — | — | |
| TAR-ViTPoseBackbone=ViT-S, Bounding box source=ground-truth ones, Temporal span T=22026.03 | 82.9 | 85.6 | 88.5 | 84.2 | 79.2 | 80.8 | 81.9 | 78.7 | — | — | — | — | — | — | |
| GNN-based pose dynamics learningfiltering=without filtering, joints_counted=all2021.06 | 82.7 | 85.1 | 87.7 | 85.3 | 80 | 81.1 | 81.6 | 77.2 | — | — | — | — | — | — | |
| DCPose + HANetBase method=DCPose2022.11 | 82.3 | 86.1 | 88.5 | 84.1 | 78.7 | 79 | 80.3 | 77.4 | — | — | — | — | — | — | |
| FAMI-Pose2022.03 | 82.2 | 85.5 | 87.7 | 84.2 | 79.2 | 81.4 | 81.1 | 74.9 | — | — | — | — | — | — | |
| FAMI-Pose2023.03 | 82.2 | 85.5 | 87.7 | 84.2 | 79.2 | 81.4 | 81.1 | 74.9 | — | — | — | — | — | — | |
| FAMIPoseBackbone=HRNet-W48, Method category=heatmap-based2024.03 | 82.2 | 85.5 | 87.7 | 84.2 | 79.2 | 81.4 | 81.1 | 74.9 | — | — | — | — | — | — | |
| FAMI-Pose2025.01 | 82.2 | 85.5 | 87.8 | 84.2 | 79.2 | 81.4 | 81.1 | 74.9 | — | — | — | — | — | — | |
| FAMI-PoseBackbone=HRNet-W48, Bounding box source=Faster R-CNN detector [28], Temporal span T=22026.03 | 82.2 | 85.5 | 87.7 | 84.2 | 79.2 | 81.4 | 81.1 | 74.9 | — | — | — | — | — | — | |
| LightTrackBackbone=CPN101, Image Size=384 x 288, #Params=46.3M, GFLOPS=22.92021.05 | 82.1 | 87.9 | 87.7 | 83.5 | 75.8 | 79.1 | 80.4 | 77 | — | — | — | — | — | — | |
| DSTABackbone=HRNet-W48, Method category=regression-based, Temporal span T=22024.03 | 82.1 | 86.2 | 88.6 | 84.2 | 78.5 | 82 | 79.2 | 73.7 | — | — | — | — | — | — | |
| TAR-ViTPoseBackbone=ViT-B, Bounding box source=Faster R-CNN detector [28], Temporal span T=22026.03 | 82.1 | 84.3 | 87.8 | 83.6 | 79.4 | 79.3 | 82.8 | 76.5 | — | — | — | — | — | — | |
| SBLBackbone=ResNet-50, Image Size=256 x 192, #Params=34.0M, GFLOPS=8.992021.05 | 81.6 | 86.7 | 88.1 | 83 | 75.7 | 80.8 | 80.4 | 74.2 | — | — | — | — | — | — | |
| T-VIPNASBackbone=ResNet-50, Image Size=256 x 192, #Params=3.9M, GFLOPS=0.822021.05 | 81.6 | 87.7 | 88.2 | 82.6 | 74.7 | 79.3 | 79.8 | 75.4 | — | — | — | — | — | — | |
| KeyTrackfiltering=without filtering, joints_counted=all2021.06 | 81.6 | 84.1 | 87.2 | 85.3 | 79.2 | 77.1 | 80.6 | 76.5 | — | — | — | — | — | — | |
| 3D Video HRNetApproach category=Top-down2020.03 | 81.5 | 84.9 | 87.4 | 84.8 | 79.2 | 77.6 | 79.7 | 75.3 | — | — | — | — | — | — | |
| CombDetfiltering=without filtering, joints_counted=all2021.06 | 81.5 | 84.9 | 87.4 | 84.8 | 79.2 | 77.6 | 79.7 | 75.3 | — | — | — | — | — | — | |
| DetTrack2023.03 | 81.5 | 84.9 | 87.4 | 84.8 | 79.2 | 77.6 | 79.7 | 75.3 | — | — | — | — | — | — | |
| DetTrackBackbone=HRNet-W48, Method category=heatmap-based2024.03 | 81.5 | 84.9 | 87.4 | 84.8 | 79.2 | 77.6 | 79.7 | 75.3 | — | — | — | — | — | — | |
| DetTrack2025.01 | 81.5 | 84.9 | 87.4 | 84.8 | 79.2 | 77.6 | 79.7 | 75.3 | — | — | — | — | — | — | |
| DetTrackBackbone=HRNet-W48, Bounding box source=unspecified sources or ground-truth ones2026.03 | 81.5 | 84.9 | 87.4 | 84.8 | 79.2 | 77.6 | 79.7 | 75.3 | — | — | — | — | — | — | |
| S-VIPNASBackbone=MobileNet-V3, Image Size=256 x 192, #Params=5.4M, GFLOPS=0.692021.05 | 81.1 | 87.8 | 88 | 82.3 | 74.1 | 78.8 | 79.1 | 74 | — | — | — | — | — | — | |
| PT_CPN++Approach category=Top-down2020.03 | 80.9 | 82.4 | 88.8 | 86.2 | 79.4 | 72 | 80.6 | 76.2 | — | — | — | — | — | — | |
| DCPoseBackbone=HRNet-W482021.03 | 80.9 | 84 | 86.6 | 82.7 | 78 | 80.4 | 79.3 | 73.8 | — | — | — | — | — | — | |
| PT_CPN++filtering=without filtering, joints_counted=all2021.06 | 80.9 | 82.4 | 88.8 | 86.2 | 79.4 | 72 | 80.6 | 76.2 | — | — | — | — | — | — | |
| DCPose2022.03 | 80.9 | 84 | 86.6 | 82.7 | 78 | 80.4 | 79.3 | 73.8 | — | — | — | — | — | — | |
| DCPose2022.11 | 80.9 | 84 | 86.6 | 82.7 | 78 | 80.4 | 79.3 | 73.8 | — | — | — | — | — | — | |
| DCPose2023.03 | 80.9 | 84 | 86.6 | 82.7 | 78 | 80.4 | 79.3 | 73.8 | — | — | — | — | — | — | |
| PT-CPN++Backbone=CPN [4], Method category=heatmap-based2024.03 | 80.9 | 82.4 | 88.8 | 86.2 | 79.4 | 72 | 80.6 | 76.2 | — | — | — | — | — | — | |
| DCPoseBackbone=HRNet-W48, Method category=heatmap-based2024.03 | 80.9 | 84 | 86.6 | 82.7 | 78 | 80.4 | 79.3 | 73.8 | — | — | — | — | — | — | |
| DCPose2025.01 | 80.9 | 84 | 86.6 | 82.7 | 78 | 80.4 | 79.3 | 73.8 | — | — | — | — | — | — | |
| DCPoseBackbone=HRNet-W48, Bounding box source=Faster R-CNN detector [28]2026.03 | 80.9 | 84 | 86.6 | 82.7 | 78 | 80.4 | 79.3 | 73.8 | — | — | — | — | — | — | |
| SBLBackbone=EfficientNetB0, Image Size=256 x 192, #Params=14.9M, GFLOPS=5.052021.05 | 80.7 | 87.3 | 87.9 | 82 | 73.4 | 79 | 79.2 | 72.9 | — | — | — | — | — | — | |
| PT-CPN++2023.03 | 80.6 | 82.4 | 88.8 | 86.2 | 79.4 | 75 | 76.7 | 72 | — | — | — | — | — | — | |
| PoseWarperTemporal Pose Aggregation=Enabled2019.06 | 79.7 | 79.9 | 86.3 | 82.4 | 77.5 | 79.8 | 78.8 | 73.2 | — | — | — | — | — | — | |
| PoseWarper2021.03 | 79.7 | 79.9 | 86.3 | 82.4 | 77.5 | 79.8 | 78.8 | 73.2 | — | — | — | — | — | — | |
| PoseWarper2022.03 | 79.7 | 79.9 | 86.3 | 82.4 | 77.5 | 79.8 | 78.8 | 73.2 | — | — | — | — | — | — | |
| PoseWarper2022.11 | 79.7 | 79.9 | 86.3 | 82.4 | 77.5 | 79.8 | 78.8 | 73.2 | — | — | — | — | — | — | |
| PoseWarper2023.03 | 79.7 | 79.9 | 86.3 | 82.4 | 77.5 | 79.8 | 78.8 | 73.2 | — | — | — | — | — | — | |
| PoseWarp.Backbone=HRNet-W48, Method category=heatmap-based2024.03 | 79.7 | 79.9 | 86.3 | 82.4 | 77.5 | 79.8 | 78.8 | 73.2 | — | — | — | — | — | — | |
| PoseWarper2025.01 | 79.7 | 79.9 | 86.3 | 82.4 | 77.5 | 79.8 | 78.8 | 73.2 | — | — | — | — | — | — | |
| PoseWarperBackbone=HRNet-W48, Bounding box source=Faster R-CNN detector [28]2026.03 | 79.7 | 79.9 | 86.3 | 82.4 | 77.5 | 79.8 | 78.8 | 73.2 | — | — | — | — | — | — | |
| PAVE-NetBackbone=Swin-L, Bounding box source=Faster R-CNN detector [28], End-to-end approach=true2026.03 | 79.7 | 84.9 | 87.5 | 81.7 | 74.5 | 77.9 | 76.9 | 72.4 | — | — | — | — | — | — | |
| SBLBackbone=ResNet-18, Image Size=256 x 192, #Params=15.3M, GFLOPS=5.792021.05 | 79.6 | 86.5 | 86.9 | 80.8 | 71.5 | 79.3 | 77.9 | 70.6 | — | — | — | — | — | — | |
| DSTABackbone=ResNet-152, Method category=regression-based, Temporal span T=22024.03 | 79.6 | 85.2 | 87.1 | 80.5 | 74.4 | 79.6 | 78 | 69.7 | — | — | — | — | — | — | |
| TAR-ViTPoseBackbone=ViT-S, Bounding box source=Faster R-CNN detector [28], Temporal span T=22026.03 | 79.3 | 82.8 | 85.6 | 81.5 | 75.2 | 78 | 78.1 | 72.1 | — | — | — | — | — | — | |
| T-VIPNASBackbone=MobileNet-V3, Image Size=256 x 192, #Params=2.5M, GFLOPS=0.372021.05 | 78.2 | 87.3 | 85.6 | 78.9 | 70.3 | 75.7 | 75 | 70.1 | — | — | — | — | — | — | |
| SBLBackbone=MobileNet-V3, Image Size=256 x 192, #Params=5.5M, GFLOPS=4.12021.05 | 78.1 | 86.4 | 85.9 | 78.8 | 69.6 | 76.5 | 76.1 | 69.1 | — | — | — | — | — | — | |
| GNN-based pose dynamics learningfiltering low confidence joints=true2021.06 | 77.9 | 80.6 | 84.5 | 80.6 | 74.4 | 75 | 76.7 | 71.9 | — | — | — | — | — | — | |
| Dynamic-GNN2022.03 | 77.9 | 80.6 | 84.5 | 80.6 | 74.4 | 75 | 76.7 | 71.8 | — | — | — | — | — | — | |
| Dynamic-GNN2023.03 | 77.9 | 80.6 | 84.5 | 80.6 | 74.4 | 72.3 | 72.2 | 71.8 | — | — | — | — | — | — | |
| Dyn.-GNNBackbone=HRNet-W48, Method category=heatmap-based2024.03 | 77.9 | 80.6 | 84.5 | 80.6 | 74.4 | 75 | 76.7 | 71.8 | — | — | — | — | — | — | |
| Dyn.-GNNBackbone=HRNet-W48, Bounding box source=unspecified sources or ground-truth ones2026.03 | 77.9 | 80.6 | 84.5 | 80.6 | 74.4 | 75 | 76.7 | 71.8 | — | — | — | — | — | — | |
| PGPT2022.03 | 76.8 | 77.5 | 82.5 | 80.5 | 74.7 | 76.2 | 76.6 | 72.2 | — | — | — | — | — | — | |
| PGPT2023.03 | 76.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| TKMRNetfiltering low confidence joints=true2021.06 | 76.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| PoseFixInput Pose=Simple [30]2018.12 | 75 | 79 | 81.6 | 76.4 | 69.7 | 75.2 | 74.3 | 67 | — | — | — | — | — | — | |
| MDPN2019.06 | 75 | 75.4 | 81.2 | 79 | 74.1 | 72.4 | 73 | 69.9 | — | — | — | — | — | — | |
| MDPN2021.03 | 75 | 75.4 | 81.2 | 79 | 74.1 | 72.4 | 73 | 69.9 | — | — | — | — | — | — | |
| MDPN2022.03 | 75 | 75.4 | 81.2 | 79 | 74.1 | 72.4 | 73 | 69.9 | — | — | — | — | — | — | |
| MDPN2023.03 | 75 | 75.4 | 81.2 | 79 | 74.1 | 72.4 | 73 | 69.9 | — | — | — | — | — | — | |
| MDPNBackbone=ResNet-152, Method category=heatmap-based2024.03 | 75 | 75.4 | 81.2 | 79 | 74.1 | 72.4 | 73 | 69.9 | — | — | — | — | — | — | |
| TML++Approach category=Bottom-up2020.03 | 74.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| TML++filtering low confidence joints=true2021.06 | 74.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| TML++2022.03 | 74.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| TML++2023.03 | 74.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LightTrackBackbone=MobileNetV1, Image Size=384 x 288, #Params=15.8M, GFLOPS=11.32021.05 | 73.4 | 85.2 | 81.7 | 74.7 | 62.9 | 72.9 | 69.4 | 61.4 | — | — | — | — | — | — | |
| AlphaPose2019.06 | 71.9 | 63.9 | 78.7 | 77.4 | 71 | 73.7 | 73 | 69.7 | — | — | — | — | — | — | |
| AlphaPose2021.03 | 71.9 | 63.9 | 78.7 | 77.4 | 71 | 73.7 | 73 | 69.7 | — | — | — | — | — | — | |
| AlphaPose2022.03 | 71.9 | 63.9 | 78.7 | 77.4 | 71 | 73.7 | 73 | 69.7 | — | — | — | — | — | — | |
| AlphaPose2023.03 | 71.9 | 63.9 | 78.7 | 77.4 | 71 | 73.7 | 73 | 69.7 | — | — | — | — | — | — | |
| AlphaPoseBackbone=ResNet-50, Method category=heatmap-based2024.03 | 71.9 | 63.9 | 78.7 | 77.4 | 71 | 73.7 | 73 | 69.7 | — | — | — | — | — | — | |
| Simple2018.12 | 70.4 | 74.4 | 76.9 | 72.2 | 65.2 | 69.2 | 70 | 62.9 | — | — | — | — | — | — | |
| STAFApproach category=Bottom-up2020.03 | 70.4 | — | — | 64.7 | — | — | 62 | — | — | — | — | — | — | — | |
| STAFfiltering low confidence joints=true2021.06 | 70.4 | — | — | — | — | 64.7 | — | 62 | — | — | — | — | — | — | |
| STAF2022.03 | 70.4 | 68.2 | 78.5 | 69.5 | 62.5 | 71.9 | 69.7 | 63.9 | — | — | — | — | — | — | |
| STAF2023.03 | 70.4 | — | — | — | 64.7 | — | — | 62 | — | — | — | — | — | — | |
| SBLBackbone=ShuffleNetV2, Image Size=256 x 192, #Params=4.7M, GFLOPS=4.212021.05 | 63.4 | 84.5 | 83 | 74.4 | 63.1 | 74.4 | 70.7 | 63.4 | — | — | — | — | — | — | |
| CCMBackbone=ResNet-50, Evaluation Setting=Zero-shot transfer (trained on COCO)2020.02 | — | — | — | — | — | — | — | — | 71.3 | 87.1 | 76.5 | 42 | 75.5 | 74.1 | |
| CCMBackbone=HRNet-w32, Evaluation Setting=Zero-shot transfer (trained on COCO)2020.02 | — | — | — | — | — | — | — | — | 73.8 | 87.7 | 79 | 44.8 | 78.2 | 76.2 | |
| EvoPose2D-MPT (Pre-training)=Y, Input Size=384x288, Params (M)=7.34, FLOPs (G)=5.59, Size (MB)=30.7, FPS (GPU)=35.82020.11 | — | — | — | — | — | — | — | — | 60.4 | 87.9 | 71.8 | 36.1 | 64 | 64.3 |