Learning Dynamical Human-Joint Affinity for 3D Pose Estimation in Videos

Benchmark Model Rank Results
3d-human-pose-estimation-on-humaneva-iDG-Net (T=4)Mean Reconstruction Error (mm): 19.5
3d-human-pose-estimation-on-mpi-inf-3dhpDG-Net (T=4)MPJPE: 76AUC: 53.8
pose-estimation-on-leeds-sports-posesDG-Net (T=4)PCK: 87.5%