PoseFormerV2: Exploring Frequency Domain for Efficient and Robust 3D Human Pose Estimation

Benchmark Model Rank Results
classification-on-full-body-parkinsonsPoseFormerV2#1F1-score (weighted): 0.59
3d-human-pose-estimation-on-mpi-inf-3dhpPoseFormerV2 (T=81)#14MPJPE: 27.8AUC: 78.8PCK: 97.9
monocular-3d-human-pose-estimation-on-human3-6mPoseFormerV2 (f=27, n=27, sequence=243)#23Average MPJPE (mm): 45.2
3d-human-pose-estimation-on-human36mPoseFormerV2 (f=27, n=27, sequence=243)#72PA-MPJPE: 35.6