3D human pose estimation in video with temporal convolutions and semi-supervised training

Benchmark Model Rank Results
3d-human-pose-estimation-on-human36mVideoPose3D (T=243)#32Average MPJPE (mm): 46.8Using 2D ground-truth joints: No
3d-human-pose-estimation-on-human36mVideoPose3D (T=1)#51Average MPJPE (mm): 51.8Using 2D ground-truth joints: No
monocular-3d-human-pose-estimation-on-human3VideoPose3D (T=243)#17Average MPJPE (mm): 46.8Use Video Sequence: YesFrames Needed: 243
weakly-supervised-3d-human-pose-estimation-on-human36mPavllo et al.#10Average MPJPE (mm): 64.7Number of Views: 13D Annotations: S1
weakly-supervised-3d-human-pose-estimation-on-human36mVideoPose3D (T=243)#22Number of Frames Per View: 243