Surface Normal Estimation on Sintel Outdoor
22.4Accuracy (11.25° Threshold)Lotus-D
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Lotus-DTraining Data=59K, Zero-shot=true2025.09 | 22.4 | 32.3 | — | — | |
| Diffusion-E2E-FTTraining Data=74K, Zero-shot=true2025.09 | 22.3 | 33.5 | — | — | |
| FE2ETraining Data=71K, Zero-shot=true2025.09 | 22.3 | 31.2 | — | — | |
| DSINETraining Data=160K, Zero-shot=true2025.09 | 21.5 | 34.9 | — | — | |
| DSINETraining Data=160K, Zero-shot=true2026.06 | 21.5 | — | 34.9 | 52.7 | |
| Lotus-GTraining Data=59K, Zero-shot=true2025.09 | 21 | 33.6 | — | — | |
| UNIGPTraining Data=59K, Zero-shot=true2026.06 | 20.1 | — | 35 | 55.1 | |
| GenPerceptTraining Data=74K, Zero-shot=true2025.09 | 16.2 | 37.6 | — | — | |
| Omnidata V2Training Data=12.2M, Zero-shot=true2026.06 | 14.7 | — | 40.5 | 43.5 | |
| StableNormalTraining Data=250K, Zero-shot=true2025.09 | 14.1 | 36.7 | — | — | |
| StableNormalTraining Data=250K, Zero-shot=true2026.06 | 14.1 | — | 36.7 | 50.7 | |
| GeoWizardTraining Data=280K, Zero-shot=true2025.09 | 12.3 | 40.3 | — | — | |
| GeoWizardTraining Data=280K, Zero-shot=true2026.06 | 12.3 | — | 40.3 | 43.5 | |
| OmnidataTraining Data=12.2M, Zero-shot=true2026.06 | 11.4 | — | 41.5 | 42 |