Depth Estimation on NYU v2
0.245RMSEFull tuning
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Full tuningBackbone=DINOv22026.05 | 0.245 | — | 96.8 | 0.068 | — | — | — | — | |
| SIGMABackbone=DINOv22026.05 | 0.259 | — | 96.3 | 0.072 | — | — | — | — | |
| LorandBackbone=DINOv22026.05 | 0.281 | — | 95.4 | 0.075 | — | — | — | — | |
| Partial-1Backbone=DINOv22026.05 | 0.282 | — | 95.2 | 0.076 | — | — | — | — | |
| LoRABackbone=DINOv22026.05 | 0.284 | — | 94.9 | 0.076 | — | — | — | — | |
| VFM-AdapterBackbone=DINOv22026.05 | 0.284 | — | 95.2 | 0.076 | — | — | — | — | |
| AdapterBackbone=DINOv22026.05 | 0.286 | — | 94.8 | 0.078 | — | — | — | — | |
| AdaptformerBackbone=DINOv22026.05 | 0.287 | — | 94.7 | 0.078 | — | — | — | — | |
| BitFitBackbone=DINOv22026.05 | 0.29 | — | 94.5 | 0.08 | — | — | — | — | |
| FixedBackbone=DINOv22026.05 | 0.292 | — | 94.2 | 0.082 | — | — | — | — | |
| Full tuningBackbone=SigLIP22026.05 | 0.294 | — | 94.7 | 0.078 | — | — | — | — | |
| V-JEPA 2.1 ViT-GParam.=2B, Patch Size=16, Evaluation Protocol=Linear probe, Input Resolution=Default2026.03 | 0.307 | — | — | — | — | — | — | — | |
| SIGMABackbone=SigLIP22026.05 | 0.307 | — | 94.1 | 0.083 | — | — | — | — | |
| DINOv3Param.=7B, Patch Size=16, Evaluation Protocol=Linear probe, Input Resolution=Default2026.03 | 0.309 | — | — | — | — | — | — | — | |
| LingBot-VisionSize=L2026.07 | 0.31 | — | — | — | — | — | — | — | |
| LoRABackbone=SigLIP22026.05 | 0.317 | — | 93.4 | 0.087 | — | — | — | — | |
| LorandBackbone=SigLIP22026.05 | 0.318 | — | 93.4 | 0.088 | — | — | — | — | |
| Full tuningBackbone=SAM2026.05 | 0.321 | — | 92.5 | 0.088 | — | — | — | — | |
| BitFitBackbone=SigLIP22026.05 | 0.321 | — | 93.1 | 0.089 | — | — | — | — | |
| Partial-1Backbone=SigLIP22026.05 | 0.321 | — | 93.1 | 0.089 | — | — | — | — | |
| AdapterBackbone=SigLIP22026.05 | 0.321 | — | 92.9 | 0.089 | — | — | — | — | |
| AdaptformerBackbone=SigLIP22026.05 | 0.321 | — | 93.1 | 0.089 | — | — | — | — | |
| FixedBackbone=SigLIP22026.05 | 0.322 | — | 93 | 0.089 | — | — | — | — | |
| VFM-AdapterBackbone=SigLIP22026.05 | 0.323 | — | 93.1 | 0.089 | — | — | — | — | |
| SIGMABackbone=SAM2026.05 | 0.334 | — | 91.8 | 0.092 | — | — | — | — | |
| VECA-LResolution=5122026.05 | 0.3379 | — | — | — | — | — | — | — | |
| VECA-LResolution=7682026.05 | 0.3382 | — | — | — | — | — | — | — | |
| LingBot-VisionSize=B2026.07 | 0.339 | — | — | — | — | — | — | — | |
| AM-RADIOv2.5Param.=1B, Patch Size=14, Evaluation Protocol=Linear probe, Input Resolution=Default2026.03 | 0.34 | — | — | — | — | — | — | — | |
| VECA-LResolution=3842026.05 | 0.3409 | — | — | — | — | — | — | — | |
| VFM-AdapterBackbone=SAM2026.05 | 0.343 | — | 91.7 | 0.094 | — | — | — | — | |
| LoRABackbone=SAM2026.05 | 0.344 | — | 91.2 | 0.098 | — | — | — | — | |
| LorandBackbone=SAM2026.05 | 0.344 | — | 91.5 | 0.096 | — | — | — | — | |
| DINOv3-LResolution=7682026.05 | 0.3445 | — | — | — | — | — | — | — | |
| DINOv3-LResolution=5122026.05 | 0.3446 | — | — | — | — | — | — | — | |
| Partial-1Backbone=SAM2026.05 | 0.345 | — | 90.8 | 0.102 | — | — | — | — | |
| AdaptformerBackbone=SAM2026.05 | 0.345 | — | 91.2 | 0.098 | — | — | — | — | |
| DINOv3-LResolution=3842026.05 | 0.346 | — | — | — | — | — | — | — | |
| AdapterBackbone=SAM2026.05 | 0.346 | — | 91 | 0.099 | — | — | — | — | |
| V-JEPA2.1Size=L2026.07 | 0.346 | — | — | — | — | — | — | — | |
| BitFitBackbone=SAM2026.05 | 0.347 | — | 90.2 | 0.108 | — | — | — | — | |
| V-JEPA 2.1 ViT-gParam.=1B, Patch Size=16, Evaluation Protocol=Linear probe, Input Resolution=Default2026.03 | 0.35 | — | — | — | — | — | — | — | |
| FixedBackbone=SAM2026.05 | 0.351 | — | 89.5 | 0.115 | — | — | — | — | |
| DINOv3Size=L2026.07 | 0.351 | — | — | — | — | — | — | — | |
| DINOv3 ViT-H+Param.=0.8B, Patch Size=16, Evaluation Protocol=Linear probe, Input Resolution=Default2026.03 | 0.352 | — | — | — | — | — | — | — | |
| VECA-LResolution=2562026.05 | 0.3546 | — | — | — | — | — | — | — | |
| PEspatialParam.=2B, Patch Size=14, Evaluation Protocol=Linear probe, Input Resolution=Default2026.03 | 0.362 | — | — | — | — | — | — | — | |
| DINOv3-LResolution=2562026.05 | 0.3663 | — | — | — | — | — | — | — | |
| DINOv3-BResolution=7682026.05 | 0.3665 | — | — | — | — | — | — | — | |
| DINOv3-BResolution=5122026.05 | 0.3684 | — | — | — | — | — | — | — | |
| DINOv3-B/16Protocol=Linear probing, Backbone=B/162026.05 | 0.3684 | — | — | — | — | — | — | — | |
| VECA-BResolution=7682026.05 | 0.3704 | — | — | — | — | — | — | — | |
| VECA-BResolution=5122026.05 | 0.3705 | — | — | — | — | — | — | — | |
| VECA-B/16Protocol=Linear probing, Backbone=B/16, Core budget (C)=642026.05 | 0.3705 | — | — | — | — | — | — | — | |
| DINOv3Size=B2026.07 | 0.371 | — | — | — | — | — | — | — | |
| DINOv2Param.=1B, Patch Size=14, Evaluation Protocol=Linear probe, Input Resolution=Default2026.03 | 0.372 | — | — | — | — | — | — | — | |
| DINOv3-BResolution=3842026.05 | 0.3727 | — | — | — | — | — | — | — | |
| DINOv3-ViT-BBackbone=ViT-B2026.03 | 0.373 | — | — | — | — | — | — | — | |
| DUNE-BBackbone=ViT-B2026.03 | 0.375 | — | — | — | — | — | — | — | |
| VECA-BResolution=3842026.05 | 0.3768 | — | — | — | — | — | — | — | |
| AM-RADIOv2.5-B/16Protocol=Linear probing, Backbone=B/162026.05 | 0.3792 | — | — | — | — | — | — | — | |
| M2H-MX-Lvariant=L2026.03 | 0.38 | — | — | — | — | — | — | — | |
| M2H-MXVariant=Large2026.05 | 0.38 | — | — | — | — | — | — | — | |
| LingBot-VisionSize=S2026.07 | 0.383 | — | — | — | — | — | — | — | |
| PEspatial-BBackbone=ViT-B2026.03 | 0.389 | — | — | — | — | — | — | — | |
| DINOv3-BResolution=2562026.05 | 0.3897 | — | — | — | — | — | — | — | |
| EUPE-ViT-BBackbone=ViT-B2026.03 | 0.391 | — | — | — | — | — | — | — | |
| VECA-BResolution=2562026.05 | 0.392 | — | — | — | — | — | — | — | |
| DINOv2-B/14Protocol=Linear probing, Backbone=B/142026.05 | 0.3949 | — | — | — | — | — | — | — | |
| DINOv2-reg-B/14Protocol=Linear probing, Backbone=B/142026.05 | 0.3953 | — | — | — | — | — | — | — | |
| DINOv3Size=S2026.07 | 0.405 | — | — | — | — | — | — | — | |
| V-JEPA2.1Size=B2026.07 | 0.41 | — | — | — | — | — | — | — | |
| DINOv2Size=L2026.07 | 0.411 | — | — | — | — | — | — | — | |
| TIGER-LBackbone=ViT-L2026.06 | 0.4115 | — | — | — | — | — | — | — | |
| DINOv3-S+Resolution=7682026.05 | 0.414 | — | — | — | — | — | — | — | |
| DINOv3-S+Resolution=5122026.05 | 0.4143 | — | — | — | — | — | — | — | |
| DINOv3-SResolution=5122026.05 | 0.4169 | — | — | — | — | — | — | — | |
| M2H-MX-Bvariant=B2026.03 | 0.417 | — | — | — | — | — | — | — | |
| M2H-MXVariant=Base2026.05 | 0.417 | — | — | — | — | — | — | — | |
| M2H2026.03 | 0.4196 | — | — | — | — | — | — | — | |
| M2H2026.05 | 0.4196 | — | — | — | — | — | — | — | |
| DINOv3-SResolution=7682026.05 | 0.4206 | — | — | — | — | — | — | — | |
| DINOv3-S+Resolution=3842026.05 | 0.4225 | — | — | — | — | — | — | — | |
| VECA-S+Resolution=7682026.05 | 0.4231 | — | — | — | — | — | — | — | |
| FrancaSize=L2026.07 | 0.425 | — | — | — | — | — | — | — | |
| DINOv2Backbone=ViT-g/14, Frozen Backbone=true2024.12 | 0.4252 | — | — | — | — | — | — | — | |
| DINOv3-SResolution=3842026.05 | 0.4252 | — | — | — | — | — | — | — | |
| VECA-S+Resolution=5122026.05 | 0.4252 | — | — | — | — | — | — | — | |
| DINOv2Size=B2026.07 | 0.429 | — | — | — | — | — | — | — | |
| VECA-SResolution=7682026.05 | 0.4298 | — | — | — | — | — | — | — | |
| SAKCategory=Conventional Multi-task Methods2026.06 | 0.4313 | — | — | — | — | — | — | — | |
| VECA-S+Resolution=3842026.05 | 0.4316 | — | — | — | — | — | — | — | |
| VECA-B/16Protocol=Linear probing, Backbone=B/16, Core budget (C)=82026.05 | 0.433 | — | — | — | — | — | — | — | |
| AM-RADIOBackbone=ViT-H/16, Frozen Backbone=true2024.12 | 0.4339 | — | — | — | — | — | — | — | |
| VECA-SResolution=5122026.05 | 0.4347 | — | — | — | — | — | — | — | |
| RADIOv2.5-BBackbone=ViT-B2026.03 | 0.435 | — | — | — | — | — | — | — | |
| RADIOv2.5Backbone=ViT-H/16, Frozen Backbone=true2024.12 | 0.4353 | — | — | — | — | — | — | — | |
| VECA-SResolution=3842026.05 | 0.436 | — | — | — | — | — | — | — | |
| DINOv2Backbone=ViT-L/14, Frozen Backbone=true2024.12 | 0.4406 | — | — | — | — | — | — | — | |
| DINOv2Size=S2026.07 | 0.447 | — | — | — | — | — | — | — |