3D Human Pose and Shape Estimation on MPI-INF-3DHP (test)
83.6MPJPEMAED
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| MAEDNumber of input frames=642023.11 | 83.6 | 56.2 | — | — | — | |
| MAEDInput=video, Training Data=w/ 3DPW2021.09 | 83.6 | 56.2 | — | — | — | |
| MAEDInput=video, Training Data=w/o 3DPW2021.09 | 85.1 | 56.5 | — | — | — | |
| HybrIK-TransformerBackbone=HrNet-482023.02 | 85.2 | — | — | 89.3 | 49 | |
| TARNumber of input frames=9, Backbone=HRNet2023.11 | 85.9 | 60.5 | 9.2 | — | — | |
| HybrIK-TransformerBackbone=HrNet-48, Training Data=w. 3DPW2023.02 | 86.2 | — | — | 88.6 | 48.9 | |
| TARNumber of input frames=9, Backbone=ResNet502023.11 | 87.5 | 62.7 | 9.6 | — | — | |
| HybrIK-TransformerBackbone=ResNet-342023.02 | 88.3 | — | — | 88.3 | 48.4 | |
| HybrIK-TransformerBackbone=ResNet-34, Training Data=w. 3DPW2023.02 | 90.2 | — | — | 86.7 | 47.6 | |
| HybrIK (Adaptive)variant=Adaptive2020.11 | 91 | — | — | 86.2 | 42.2 | |
| HybrIK2023.02 | 91 | — | — | 86.2 | 42.2 | |
| HybrIK (Naive)variant=Naive2020.11 | 91.5 | — | — | 85.9 | 41.7 | |
| DST-VIBEInput Type=Video, Use 3DPW Train Set=true, Use Human3.6 Train Set=true2021.10 | 93.4 | 62.2 | 11.9 | — | — | |
| HybrIK (Adaptive)variant=Adaptive, trained with 3DPW=true2020.11 | 93.9 | — | — | 87.5 | 46.9 | |
| HybrIKTraining Data=w. 3DPW2023.02 | 93.9 | — | — | 87.5 | 46.9 | |
| GLoTNumber of input frames=162023.11 | 93.9 | 61.5 | 7.9 | — | — | |
| TePoseReal-time capability=true, Training dataset=3DPW train set, Number of input frames (T+1)=62022.07 | 96.2 | 63.1 | 16.7 | — | — | |
| MEVAInput Type=Video, Use 3DPW Train Set=true, Use Human3.6 Train Set=false2021.10 | 96.4 | 65.4 | 11.1 | — | — | |
| MEVAReal-time capability=false, Training dataset=3DPW train set, Number of input frames (T+1)=902022.07 | 96.4 | 65.4 | 11.1 | — | — | |
| MEVANumber of input frames=902023.11 | 96.4 | 65.4 | 11.1 | — | — | |
| MEVAInput=video, Training Data=w/ 3DPW2021.09 | 96.4 | 65.4 | — | — | — | |
| TCMRInput type=video2020.11 | 96.5 | 62.8 | 9.5 | — | — | |
| VIBEInput Type=Video, Use 3DPW Train Set=true, Use Human3.6 Train Set=true2021.10 | 96.6 | 64.6 | 27.3 | — | — | |
| VIBEInput=video, Training Data=w/ 3DPW2021.09 | 96.6 | 64.6 | — | — | — | |
| MPS-NetNumber of input frames=162023.11 | 96.7 | 62.8 | 9.6 | — | — | |
| TCMRReal-time capability=false, Training dataset=3DPW train set, Number of input frames (T+1)=162022.07 | 97.3 | 63.5 | 8.5 | — | — | |
| TCMRNumber of input frames=162023.11 | 97.6 | 63.5 | 8.5 | — | — | |
| VIBEInput type=video2020.11 | 97.7 | 63.4 | 29 | — | — | |
| VIBEInput=video, Training Data=w/o 3DPW2021.09 | 97.7 | 63.4 | — | — | — | |
| VIBEReal-time capability=true, Training dataset=3DPW train set, Number of input frames (T+1)=162022.07 | 103.9 | 68.9 | 27.3 | — | — | |
| VIBENumber of input frames=162023.11 | 103.9 | 68.9 | 27.3 | — | — | |
| DSRTraining data=w/ 3DPW train2021.10 | 104.7 | 67 | — | — | — | |
| SPINInput type=single image2020.11 | 105.2 | 67.5 | — | — | — | |
| SPINInput Type=Single image, Use Human3.6 Train Set=true2021.10 | 105.2 | 67.5 | — | — | — | |
| SPINInput Type=Video, Use Human3.6 Train Set=true2021.10 | 105.2 | 67.5 | — | — | — | |
| SPINTraining data=standard2021.10 | 105.2 | 67.5 | — | — | — | |
| SPIN2020.11 | 105.2 | — | — | 76.4 | 37.1 | |
| SPIN2023.02 | 105.2 | — | — | 76.4 | 37.1 | |
| SPINInput=image, Training Data=w/o 3DPW2021.09 | 105.2 | 67.5 | — | — | — | |
| DSRTraining data=standard2021.10 | 105.3 | 66.7 | — | — | — | |
| HMRInput type=single image2020.11 | 124.2 | 89.8 | — | — | — | |
| HMRInput Type=Single image, Use Human3.6 Train Set=true2021.10 | 124.2 | 89.8 | — | — | — | |
| HMRTraining data=standard2021.10 | 124.2 | 89.8 | — | — | — | |
| HMR2020.11 | 124.2 | — | — | 72.9 | 36.5 | |
| HMR2023.02 | 124.2 | — | — | 72.9 | 36.5 | |
| HMRInput=image, Training Data=w/o 3DPW2021.09 | 124.2 | 89.8 | — | — | — | |
| EFTTraining data=standard2021.10 | — | 68 | — | — | — | |
| EFTTraining data=w/ 3DPW train2021.10 | — | 67 | — | — | — |