SRNet: Improving Generalization in 3D Human Pose Estimation with a Split-and-Recombine Approach

Benchmark Model Rank Results
3d-human-pose-estimation-on-human36mSRNet (T=243)#26Average MPJPE (mm): 44.8Using 2D ground-truth joints: No
3d-human-pose-estimation-on-human36mSRNet (T=1)#43Average MPJPE (mm): 49.9Using 2D ground-truth joints: No
3d-human-pose-estimation-on-mpi-inf-3dhpSRNET#72AUC: 43.8PCK: 77.6
monocular-3d-human-pose-estimation-on-human3SRNET#19Average MPJPE (mm): 49.9Use Video Sequence: NoFrames Needed: 1