Video Depth Estimation on Sintel 384x384 (test)
0.312Abs Rel ErrorVDPP
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| VDPPMethod Category=Post-processing, Base Model/Backbone=DAv2-B, Post-processing Module=Ours2026.04 | 0.312 | 73.1 | 1.756 | 112 | 8.9 | 7.1 | 1.8 | |
| VDPPMethod Category=Post-processing, Base Model/Backbone=DAv2-L, Post-processing Module=Ours2026.04 | 0.319 | 71.3 | 1.704 | 66 | 15.1 | 13.3 | 1.8 | |
| VDA-LMethod Category=End-to-end2026.04 | 0.324 | 72.9 | 1.424 | 18 | 55.5 | — | — | |
| DAv2-LMethod Category=Image to Depth2026.04 | 0.352 | 67.1 | 1.711 | 75 | 13.3 | 13.3 | — | |
| VDA-SMethod Category=End-to-end2026.04 | 0.355 | 68.1 | 1.76 | 65.6 | 15.3 | — | — | |
| DAv2-BMethod Category=Image to Depth2026.04 | 0.358 | 67.8 | 1.809 | 141 | 7.1 | 7.1 | — | |
| DAv2-L + NVDSMethod Category=Post-processing, Base Model/Backbone=DAv2-L, Post-processing Module=NVDS2026.04 | 0.384 | 67.4 | 2.18 | 4 | 230.5 | 13.3 | 217.2 | |
| DAv2-B + NVDSMethod Category=Post-processing, Base Model/Backbone=DAv2-B, Post-processing Module=NVDS2026.04 | 0.387 | 66.8 | 2.304 | 4 | 224.3 | 7.1 | 217.2 | |
| DPT-L + NVDSMethod Category=Post-processing, Base Model/Backbone=DPT-L, Post-processing Module=NVDS2026.04 | 0.471 | 61.5 | 2.407 | 4 | 225 | 7.7 | 217.2 | |
| DPT-LMethod Category=Image to Depth2026.04 | 0.548 | 61 | 2.545 | 129 | 7.7 | 7.7 | — | |
| VDPPMethod Category=Post-processing, Base Model/Backbone=DPT-L, Post-processing Module=Ours2026.04 | 0.559 | 63.3 | 2.207 | 104 | 9.6 | 7.7 | 1.8 |