Metric Depth Estimation on HyperSim
0.232AbsRelUniK3D-V-L
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| UniK3D-V-LBackbone=ViT-Large, Train Images=8M, Zero-shot=true2025.06 | 0.232 | 66 | 3.75 | |
| Unidepth-V-LBackbone=ViT-Large, Train Images=3M, Zero-shot=true2025.06 | 0.294 | 54.5 | 4 | |
| UniK3D-V-SBackbone=ViT-Small, Train Images=8M, Zero-shot=true2025.06 | 0.349 | 52.5 | 6 | |
| TR2MBackbone=ViT-Small, Train Images=102K, Zero-shot=true2025.06 | 0.357 | 36.1 | 2.4 | |
| Depth AnythingZero-shot=true, Training set=NYUv22024.01 | 0.363 | 36.1 | — | |
| DA SingleBackbone=ViT-Large, Train Images=-, Zero-shot=true2025.06 | 0.363 | 36.1 | 2.8 | |
| ZoeDepthZero-shot=true, Training set=NYUv22024.01 | 0.407 | 30.2 | — | |
| ZoeDepthBackbone=BeiT384-Large, Train Images=-, Zero-shot=true2025.06 | 0.407 | 30.2 | 4.7 | |
| DA MixBackbone=ViT-Large, Train Images=102K, Zero-shot=true2025.06 | 0.412 | 29.6 | 5 | |
| RSABackbone=ViT-Small, Train Images=102K, Zero-shot=true2025.06 | 0.461 | 23 | 5.8 | |
| Marigold2024.11 | 13.5 | 87.5 | — | |
| DiT-based Diffusion for Perceptual TasksBackbone=DiT-XL2024.11 | 13.6 | 87.6 | — |