Depth Estimation on NYU v2 (val)
0.201RMSEUniDepth-V
Evaluation Results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| UniDepth-Vzero_shot=true, backbone=ViT [12]2024.03 | 0.201 | 98.4 | 99.7 | 99.9 | 5.78 | 0.024 | 85.9 | 0.092 | 5.27 | — | — | |
| UniDepth-V+zero_shot=true, backbone=ViT [12], predicted_intrinsics_conditioning=true2024.03 | 0.205 | 98.3 | 99.7 | 99.9 | 6.04 | 0.025 | 85.5 | 0.094 | 5.28 | — | — | |
| UniDepth-Czero_shot=true, backbone=ConvNext [33]2024.03 | 0.231 | 97.2 | 99.6 | 99.9 | 6.22 | 0.026 | 84.4 | 0.101 | 6.39 | — | — | |
| UniDepth-C+zero_shot=true, backbone=ConvNext [33], predicted_intrinsics_conditioning=true2024.03 | 0.232 | 97.2 | 99.6 | 99.9 | 6.33 | 0.027 | 84.1 | 0.103 | 6.4 | — | — | |
| ZoeDepthtrained_on=NYU2024.03 | 0.278 | 95.2 | 99.5 | 99.8 | 7.7 | 0.033 | 80.1 | 0.125 | 7.19 | — | — | |
| iDisctrained_on=NYU2024.03 | 0.313 | 93.8 | 99.2 | 99.8 | 8.61 | 0.037 | 78.2 | 0.133 | 8.85 | — | — | |
| DINOv3Model=ViT-L, Size=10242026.01 | 0.3235 | — | — | — | — | — | — | — | — | — | — | |
| C-RADIOv3Model=ViT-L, Size=10242026.01 | 0.3239 | — | — | — | — | — | — | — | — | — | — | |
| HUVRModel=ViT-B, Size=7682026.01 | 0.3263 | — | — | — | — | — | — | — | — | — | — | |
| HUVRModel=ViT-L, Size=10242026.01 | 0.3287 | — | — | — | — | — | — | — | — | — | — | |
| DDPdiffusion steps=32023.03 | 0.329 | 92.1 | 99 | 99.8 | 0.094 | 0.04 | — | — | — | — | — | |
| BinsFormerStatus=Fine-tuned SOTA2022.06 | 0.33 | — | — | — | — | — | — | — | — | — | — | |
| DINOv3Model=ViT-B, Size=7682026.01 | 0.3305 | — | — | — | — | — | — | — | — | — | — | |
| NeWCRFtrained_on=NYU2024.03 | 0.333 | 92.1 | 99.1 | 99.8 | 9.56 | 0.04 | 75.8 | 0.147 | 9.16 | — | — | |
| C-RADIOv3Model=ViT-B, Size=7682026.01 | 0.3359 | — | — | — | — | — | — | — | — | — | — | |
| Metric3Dzero_shot=true2024.03 | 0.337 | 92.6 | 97.9 | 99.1 | 9.38 | 0.038 | 77.8 | 0.146 | 9.13 | — | — | |
| MMMT-DInputs=RGB + D, Out=S + D, loss=GTS + GTD, ||A||=500, ||B||=250, ||v||=395, ||f1||=D, ||f2||=S2026.02 | 0.337 | — | — | — | — | 0.102 | — | — | — | — | — | |
| DepthFormer2023.03 | 0.339 | 92.1 | 98.9 | 99.8 | 0.096 | 0.041 | — | — | — | — | — | |
| DPT-Hybrid2023.03 | 0.357 | 90.4 | 98.8 | 99.8 | 0.11 | 0.045 | — | — | — | — | — | |
| AdaBins2023.03 | 0.364 | 90.3 | 98.4 | 99.7 | 0.103 | 0.044 | — | — | — | — | — | |
| TransDepth2023.03 | 0.365 | 90 | 98.3 | 99.6 | 0.106 | 0.045 | — | — | — | — | — | |
| AdaBinstrained_on=NYU2024.03 | 0.365 | 90.1 | 98.3 | 99.6 | 10.3 | 0.044 | 74.7 | 0.156 | 10.6 | — | — | |
| UNIFIED-IO XLModel Size=XL2022.06 | 0.385 | — | — | — | — | — | — | — | — | — | — | |
| BTStrained_on=NYU2024.03 | 0.391 | 88.5 | 97.8 | 99.4 | 10.9 | 0.046 | 74 | 0.16 | 11.5 | — | — | |
| BTS2023.03 | 0.392 | 88.5 | 97.8 | 99.4 | 0.11 | 0.047 | — | — | — | — | — | |
| UNIFIED-IO LARGEModel Size=LARGE2022.06 | 0.402 | — | — | — | — | — | — | — | — | — | — | |
| DAV2023.03 | 0.412 | 88.2 | 98 | 99.6 | 0.108 | — | — | — | — | — | — | |
| M2Hchannels=2562025.10 | 0.4196 | — | — | — | — | — | — | — | — | — | — | |
| M2H-smallchannels=642025.10 | 0.4365 | — | — | — | — | — | — | — | — | — | — | |
| UVIMStatus=Unified SOTA2022.06 | 0.467 | — | — | — | — | — | — | — | — | — | — | |
| UNIFIED-IO BASEModel Size=BASE2022.06 | 0.469 | — | — | — | — | — | — | — | — | — | — | |
| MTMamba++2025.10 | 0.4818 | — | — | — | — | — | — | — | — | — | — | |
| MTMamba2025.10 | 0.5066 | — | — | — | — | — | — | — | — | — | — | |
| MLoRE2025.10 | 0.5076 | — | — | — | — | — | — | — | — | — | — | |
| DORN2023.03 | 0.509 | 82.8 | 96.5 | 99.2 | 0.115 | 0.051 | — | — | — | — | — | |
| SigLIP 2Model=ViT-L, Size=10242026.01 | 0.5107 | — | — | — | — | — | — | — | — | — | — | |
| InvPt+Bi-MTPD-C2025.10 | 0.515 | — | — | — | — | — | — | — | — | — | — | |
| TaskPrompter2025.10 | 0.5152 | — | — | — | — | — | — | — | — | — | — | |
| SwinMTL2025.10 | 0.5179 | — | — | — | — | — | — | — | — | — | — | |
| InvPT2025.10 | 0.5183 | — | — | — | — | — | — | — | — | — | — | |
| MonoIndoor2023.03 | 0.526 | 82.3 | 95.8 | 98.9 | 0.134 | — | — | — | — | — | — | |
| SigLIP 2Model=ViT-B, Size=7682026.01 | 0.5317 | — | — | — | — | — | — | — | — | — | — | |
| MQTransformer2025.10 | 0.5325 | — | — | — | — | — | — | — | — | — | — | |
| StructDepth2023.03 | 0.534 | 81.7 | 95.5 | 98.8 | 0.14 | 0.06 | — | — | — | — | — | |
| ATRC2025.10 | 0.5363 | — | — | — | — | — | — | — | — | — | — | |
| MTI-Net2025.10 | 0.5365 | — | — | — | — | — | — | — | — | — | — | |
| C-RADIOv3Model=ViT-L, Size=322026.01 | 0.543 | — | — | — | — | — | — | — | — | — | — | |
| HUVRModel=ViT-L, Size=322026.01 | 0.5726 | — | — | — | — | — | — | — | — | — | — | |
| HUVRModel=ViT-B, Size=322026.01 | 0.598 | — | — | — | — | — | — | — | — | — | — | |
| C-RADIOv3Model=ViT-B, Size=322026.01 | 0.6017 | — | — | — | — | — | — | — | — | — | — | |
| UNIFIED-IO SMALLModel Size=SMALL2022.06 | 0.649 | — | — | — | — | — | — | — | — | — | — | |
| CATSBackbone=ResNet50, Audit Protocol=shared fixed-δ auditing, Number of seeds=72026.06 | 0.6594 | — | — | — | — | — | — | — | — | — | — | |
| GradNormBackbone=ResNet50, Audit Protocol=shared fixed-δ auditing, Number of seeds=72026.06 | 0.6599 | — | — | — | — | — | — | — | — | — | — | |
| UncertaintyBackbone=ResNet50, Audit Protocol=shared fixed-δ auditing, Number of seeds=72026.06 | 0.6609 | — | — | — | — | — | — | — | — | — | — | |
| ReLiFBackbone=ResNet50, Audit Protocol=shared fixed-δ auditing, Number of seeds=72026.06 | 0.661 | — | — | — | — | — | — | — | — | — | — | |
| ERMBackbone=ResNet50, Audit Protocol=shared fixed-δ auditing, Number of seeds=72026.06 | 0.6634 | — | — | — | — | — | — | — | — | — | — | |
| DINOv3Model=ViT-L, Size=322026.01 | 0.6685 | — | — | — | — | — | — | — | — | — | — | |
| FairGradBackbone=ResNet50, Audit Protocol=shared fixed-δ auditing, Number of seeds=72026.06 | 0.6804 | — | — | — | — | — | — | — | — | — | — | |
| PCGradBackbone=ResNet50, Audit Protocol=shared fixed-δ auditing, Number of seeds=72026.06 | 0.6812 | — | — | — | — | — | — | — | — | — | — | |
| DINOv3Model=ViT-B, Size=322026.01 | 0.7056 | — | — | — | — | — | — | — | — | — | — | |
| SigLIP 2Model=ViT-L, Size=322026.01 | 0.774 | — | — | — | — | — | — | — | — | — | — | |
| SigLIP 2Model=ViT-B, Size=322026.01 | 0.8155 | — | — | — | — | — | — | — | — | — | — | |
| FeatUp (Implicit)evaluation_protocol=linear probe2024.03 | 1.04 | — | — | — | — | — | — | — | — | 92.7 | — | |
| CARAFEevaluation_protocol=linear probe2024.03 | 1.09 | — | — | — | — | — | — | — | — | 92 | — | |
| FeatUp (JBU)evaluation_protocol=linear probe2024.03 | 1.09 | — | — | — | — | — | — | — | — | 93.8 | — | |
| Resize-convevaluation_protocol=linear probe2024.03 | 1.14 | — | — | — | — | — | — | — | — | 91.7 | — | |
| Bilinearevaluation_protocol=linear probe2024.03 | 1.19 | — | — | — | — | — | — | — | — | 91 | — | |
| DIPevaluation_protocol=linear probe, inference-time optimization=true2024.03 | 1.19 | — | — | — | — | — | — | — | — | 90.7 | — | |
| SAPAevaluation_protocol=linear probe2024.03 | 1.19 | — | — | — | — | — | — | — | — | 91.7 | — | |
| Low-resevaluation_protocol=linear probe2024.03 | 1.25 | — | — | — | — | — | — | — | — | 89.4 | — | |
| Large imageevaluation_protocol=linear probe2024.03 | 2.33 | — | — | — | — | — | — | — | — | 89.6 | — | |
| Stridedevaluation_protocol=linear probe2024.03 | 2.62 | — | — | — | — | — | — | — | — | 90 | — | |
| AnyUpParams (M)=0.88, Tra. Time (hour)=10.1, FLOPs (G)=328.5, Inf. Speed (FPS)=5.5, Inf. GPU Mem. (G)=6.0, Upsampling Ratio=x16, Evaluation Mode=DINOv2-S probing2026.03 | — | 83.61 | — | — | — | — | — | — | — | — | — | |
| DiffusionMTLLabel Regime=One2025.06 | — | — | — | — | — | — | — | — | — | — | 0.5059 | |
| DiffusionMTLLabel Regime=Random2025.06 | — | — | — | — | — | — | — | — | — | — | 0.4743 | |
| DiveUpParams (M)=0.66, Tra. Time (hour)=1.5, FLOPs (G)=265.1, Inf. Speed (FPS)=17.8, Inf. GPU Mem. (G)=1.7, Upsampling Ratio=x16, Evaluation Mode=DINOv2-S probing2026.03 | — | 84.73 | — | — | — | — | — | — | — | — | — | |
| FeatUpParams (M)=0.17, Tra. Time (hour)=23.2, FLOPs (G)=83.0, Inf. Speed (FPS)=17.5, Inf. GPU Mem. (G)=2.6, Upsampling Ratio=x16, Evaluation Mode=DINOv2-S probing2026.03 | — | 82.53 | — | — | — | — | — | — | — | — | — | |
| FisherSetting=FU, Backbone=ViT-L2026.05 | — | 56.12 | — | — | — | — | — | — | — | — | — | |
| FisherSetting=PU (SEG), Backbone=ViT-L2026.05 | — | 69.14 | — | — | — | — | — | — | — | — | — | |
| FisherSetting=PU (DEP), Backbone=ViT-L2026.05 | — | 33.92 | — | — | — | — | — | — | — | — | — | |
| FisherSetting=PU (NOR), Backbone=ViT-L2026.05 | — | 68.39 | — | — | — | — | — | — | — | — | — | |
| InfluenceSetting=FU, Backbone=ViT-L2026.05 | — | 55.29 | — | — | — | — | — | — | — | — | — | |
| InfluenceSetting=PU (SEG), Backbone=ViT-L2026.05 | — | 61.93 | — | — | — | — | — | — | — | — | — | |
| InfluenceSetting=PU (DEP), Backbone=ViT-L2026.05 | — | 53.85 | — | — | — | — | — | — | — | — | — | |
| InfluenceSetting=PU (NOR), Backbone=ViT-L2026.05 | — | 67.7 | — | — | — | — | — | — | — | — | — | |
| JAFARParams (M)=0.63, Tra. Time (hour)=2.7, FLOPs (G)=366.5, Inf. Speed (FPS)=9.2, Inf. GPU Mem. (G)=5.3, Upsampling Ratio=x16, Evaluation Mode=DINOv2-S probing2026.03 | — | 83.4 | — | — | — | — | — | — | — | — | — | |
| JTRLabel Regime=One2025.06 | — | — | — | — | — | — | — | — | — | — | 0.5919 | |
| JTRLabel Regime=Random2025.06 | — | — | — | — | — | — | — | — | — | — | 0.5541 | |
| MTPSLLabel Regime=One2025.06 | — | — | — | — | — | — | — | — | — | — | 0.6088 | |
| MTPSLLabel Regime=Random2025.06 | — | — | — | — | — | — | — | — | — | — | 0.5738 | |
| NAFParams (M)=0.66, Tra. Time (hour)=1.0, FLOPs (G)=265.1, Inf. Speed (FPS)=17.8, Inf. GPU Mem. (G)=1.7, Upsampling Ratio=x16, Evaluation Mode=DINOv2-S probing2026.03 | — | 84.06 | — | — | — | — | — | — | — | — | — | |
| NegGrad+Setting=FU, Backbone=ViT-L2026.05 | — | 68.38 | — | — | — | — | — | — | — | — | — | |
| NegGrad+Setting=PU (SEG), Backbone=ViT-L2026.05 | — | 69.97 | — | — | — | — | — | — | — | — | — | |
| NegGrad+Setting=PU (DEP), Backbone=ViT-L2026.05 | — | 69.65 | — | — | — | — | — | — | — | — | — | |
| NegGrad+Setting=PU (NOR), Backbone=ViT-L2026.05 | — | 69.68 | — | — | — | — | — | — | — | — | — | |
| OriginalSetting=Ref., Backbone=ViT-L2026.05 | — | 71.3 | — | — | — | — | — | — | — | — | — | |
| OrthoGradSetting=FU, Backbone=ViT-L2026.05 | — | 67.28 | — | — | — | — | — | — | — | — | — | |
| OrthoGradSetting=PU (SEG), Backbone=ViT-L2026.05 | — | 68.92 | — | — | — | — | — | — | — | — | — | |
| OrthoGradSetting=PU (DEP), Backbone=ViT-L2026.05 | — | 67.29 | — | — | — | — | — | — | — | — | — | |
| OrthoGradSetting=PU (NOR), Backbone=ViT-L2026.05 | — | 70.25 | — | — | — | — | — | — | — | — | — |