Monocular Depth Estimation on NYU Depth v2 (Underrepresented classes)
97.5Delta 1RAD-Large
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| RAD-LargeModel variant=Large2026.02 | 97.5 | 99.5 | 99.9 | 0.063 | 0.288 | 0.04 | |
| RAD-BaseModel variant=Base2026.02 | 96.7 | 99.4 | 99.7 | 0.072 | 0.299 | 0.053 | |
| RAD-SmallModel variant=Small2026.02 | 95.3 | 98.5 | 99.1 | 0.084 | 0.321 | 0.057 | |
| UniDepthV2Evaluation protocol=zero-shot2026.02 | 92.8 | 98.1 | 99.3 | 0.092 | 0.329 | 0.039 | |
| Metric3Dv2Evaluation protocol=zero-shot2026.02 | 92.4 | 97.6 | 99.4 | 0.095 | 0.331 | 0.039 | |
| DepthAnyV2Evaluation protocol=fine-tuning2026.02 | 91.4 | 98.2 | 99.8 | 0.089 | 0.327 | 0.049 | |
| ZoeDepthEvaluation protocol=fine-tuning2026.02 | 90.9 | 98.1 | 99.6 | 0.098 | 0.334 | 0.043 | |
| DepthProEvaluation protocol=zero-shot2026.02 | 89.6 | 97.6 | 99.1 | 0.106 | 0.375 | 0.048 | |
| AdaBinsEvaluation protocol=fine-tuning2026.02 | 81 | 88.3 | 93.7 | 0.173 | 0.522 | 0.053 | |
| NewCRFEvaluation protocol=fine-tuning2026.02 | 80.3 | 89.7 | 95.4 | 0.169 | 0.497 | 0.049 |