BEDLAM Leaderboard
The following table ranks the performance of state-of-the-art SMPL-X methods on the BEDLAM test.
| Method | NMVE | NMJE | MVE | MPJPE | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| FB | B | FB | B | FB | B | F | LH/RH | FB | B | F | LH/RH | |
| BEDLAM-CLIFF-X+ | 93.4 | 61.2 | 92.5 | 60.4 | 87.8 | 56.8 | 27.3 | 31.9/33.8 | 87.0 | 57.5 | 28.0 | 28.5/31.1 |
| BEDLAM-CLIFF-X | 101.7 | 65.6 | 99.0 | 64.7 | 95.6 | 61.7 | 29.9 | 35.7/36.2 | 93.1 | 60.8 | 30.5 | 33.3/33.3 |
| PIXIE (Feng et al., 3DV 2021) | 160.0 | 107.2 | 154.8 | 103.5 | 150.4 | 100.8 | 51.4 | 47.2/50.2 | 145.6 | 97.3 | 55.4 | 43.6/46.0 |
| PyMAF-X (Zhang et al., 2022) | 172.1 | 123.6 | 167.2 | 120.1 | 161.8 | 117.4 | 50.3 | 40.5/42.6 | 157.2 | 114.1 | 51.6 | 38.2/39.7 |
| Hand4Whole (Moon et al., CVPRW 2022) | 178.8 | 119.1 | 176.2 | 117.6 | 168.1 | 112.0 | 59.7 | 52.8/55.8 | 165.7 | 110.5 | 63.7 | 50.0/52.0 |
Note: "+" denotes that the method was trained on the AGORA training set in addition to BEDLAM.