Table 10 Quantitative comparison on HO3D Dataset (best results are in bold, while second best are italics).
From: Scene flow based deep network for hand reconstruction using depth images
Sr. | Method | Input data | Multiview | Temporal | MPVPE\(\downarrow\) | MPJPE\(\downarrow\) | MPVPE-RR\(\downarrow\) | MPJPE-RR\(\downarrow\) | Accel-RR\(\downarrow\) |
|---|---|---|---|---|---|---|---|---|---|
1 | THOR-Net11 | RGB | – | – | 23.70 | 25.60 | – | – | – |
2 | METRO14 | RGB | – | – | – | – | – | 38.70 | 17.40 |
3 | PyMAF-X109 | RGB | – | Yes | – | – | – | 29.60 | \(\mathbf {9.30}\) |
4 | HandOccNet85 | RGB | – | Yes | – | – | – | 24.9 | – |
5 | POEM119 | RGB | Yes | – | 17.20 | 17.28 | 21.45 | 21.94 | – |
6 | MLP Hand123 | RGB | Yes | – | 18.69 | 18.70 | 23.28 | 23.76 | – |
7 | ShapeGraFormer134 | D | – | – | 19.40 | 19.90 | – | – | – |
8 | HandFlowNet (ours) | Point | Yes | Yes | \(\mathbf {13.79}\) | \(\mathbf {14.41}\) | \(\mathbf {12.73}\) | \(\mathbf {13.35}\) | 9.65 |
