Depth Estimation on ScanNet (test)
0.056Abs RelDeFiNe
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| DeFiNeParam (M)=30.8, Time (ms)=782022.11 | 0.056 | — | 0.019 | 0.176 | — | — | — | — | — | |
| DeepV2DTime (s)=0.692022.11 | 0.057 | — | 0.01 | 0.168 | 0.08 | — | — | — | — | |
| DeepV2DTime (s)=0.692022.11 | 0.057 | — | 0.01 | 0.168 | 0.08 | — | — | — | — | |
| GMDepthTime (s)=0.042022.11 | 0.059 | — | 0.019 | 0.179 | 0.082 | — | — | — | — | |
| GMDepthParam (M)=7.3, Time (ms)=402022.11 | 0.059 | — | 0.019 | 0.179 | 0.082 | — | — | — | — | |
| GMDepthTime (s)=0.042022.11 | 0.059 | — | 0.019 | 0.179 | 0.082 | — | — | — | — | |
| CrossFlowZero-shot=true, # Training samples=74K2024.12 | 0.068 | — | — | — | — | 94.2 | — | — | — | |
| GMDepthParam (M)=4.7, Time (ms)=172022.11 | 0.069 | — | 0.025 | 0.211 | 0.097 | — | — | — | — | |
| MarigoldZero-shot=true, # Training samples=74K2024.12 | 0.069 | — | — | — | — | 94.5 | — | — | — | |
| CSTM_imageBackbone=ConvNeXt-large, #Params=198M, Zero-shot transfer=true, Manually aligned scale and shift=true2023.07 | 0.074 | — | — | — | — | 94.2 | — | — | — | |
| CSTM_labelBackbone=ConvNeXt-large, Zero-shot transfer=true, Manually aligned scale and shift=true2023.07 | 0.074 | — | — | — | — | 94.1 | — | — | — | |
| DepthFormerParam (M)=5.4, Time (ms)=292022.11 | 0.075 | — | 0.029 | 0.23 | 0.106 | — | — | — | — | |
| OmnidataZero-shot=true, # Training samples=11.9M + 310K2024.12 | 0.075 | — | — | — | — | 93.6 | — | — | — | |
| OmnidataBackbone=ViT-base, Zero-shot transfer=true, Manually aligned scale and shift=true2023.07 | 0.077 | — | — | — | — | 93.5 | — | — | — | |
| DPT-largeBackbone=ViT-Large, Zero-shot transfer=true, Manually aligned scale and shift=true2023.07 | 0.078 | — | — | — | — | 93.8 | — | — | — | |
| Structural Multiplane ImageInput=Image pairs (0.05~0.15m translation)2023.03 | 0.079 | — | — | 0.205 | — | 94.6 | 98.9 | — | — | |
| HDNBackbone=ViT-Large, #Params=306M, Zero-shot transfer=true, Manually aligned scale and shift=true2023.07 | 0.08 | — | — | — | — | 93.9 | — | — | — | |
| HDNZero-shot=true, # Training samples=300K2024.12 | 0.08 | — | — | — | — | 93.9 | — | — | — | |
| MaGNet (full)Cap=5m2021.12 | 0.0804 | 0.1409 | — | 0.196 | 0.1084 | 93.13 | — | — | — | |
| Long et al.Cap=5m2021.12 | 0.0805 | 0.1438 | — | 0.2029 | 0.1083 | 93.33 | — | — | — | |
| MaGNet (full)Cap=10m2021.12 | 0.081 | 0.1466 | — | 0.2098 | 0.1101 | 92.98 | — | — | — | |
| Long et al.Cap=10m2021.12 | 0.0812 | 0.1505 | — | 0.2199 | 0.1104 | 93.13 | — | — | — | |
| DPTZero-shot=true, # Training samples=1.2M + 188K2024.12 | 0.082 | — | — | — | — | 93.4 | — | — | — | |
| UCS-NetCap=10m2021.12 | 0.0845 | 0.1605 | — | 0.2335 | 0.1145 | 92.22 | — | — | — | |
| PlaneMVSInput=Image pairs (0.05~0.15m translation)2023.03 | 0.088 | — | — | 0.186 | — | 92.6 | 98.8 | — | — | |
| LeReSZero-shot=true, # Training samples=300K + 54K2024.12 | 0.091 | — | — | — | — | 91.7 | — | — | — | |
| DELTASCap=10m2021.12 | 0.0915 | 0.171 | — | 0.239 | 0.1226 | 91.47 | — | — | — | |
| DELTASsequence length=3, frame gap=202020.03 | 0.0932 | 0.154 | 0.0506 | 0.2505 | 0.1426 | 92.87 | 97.67 | 98.93 | — | |
| NASCap=10m2021.12 | 0.0941 | 0.1928 | — | 0.2703 | 0.1269 | 90.09 | — | — | — | |
| LeresBackbone=ResNeXt101, Zero-shot transfer=true, Manually aligned scale and shift=true2023.07 | 0.095 | — | — | — | — | 91.2 | — | — | — | |
| DPSNetCap=10m2021.12 | 0.0986 | 0.1998 | — | 0.284 | 0.1348 | 88.8 | — | — | — | |
| NeuralRGBDCap=5m2021.12 | 0.1013 | 0.1657 | — | 0.25 | 0.1315 | 91.6 | — | — | — | |
| DPSNetfine-tuned on ScanNet=true, sequence length=3, frame gap=202020.03 | 0.1025 | 0.1675 | 0.0574 | 0.2679 | 0.1531 | 91.02 | 97.08 | 98.72 | — | |
| MVDepthNetfine-tuned on ScanNet=true, sequence length=3, frame gap=202020.03 | 0.1054 | 0.1911 | 0.097 | 0.3053 | 0.1553 | 89.52 | 97.07 | 98.95 | — | |
| NAS2021.12 | 0.107 | — | — | 0.2807 | — | — | — | — | — | |
| GPMVSNetfine-tuned on ScanNet=true, sequence length=3, frame gap=202020.03 | 0.1079 | 0.2255 | 0.096 | 0.4659 | 0.1998 | 89.05 | 95.91 | 97.89 | — | |
| DiverseDepthBackbone=ResNeXt50, #Params=25M, Zero-shot transfer=true, Manually aligned scale and shift=true2023.07 | 0.108 | — | — | — | — | 88.2 | — | — | — | |
| DiverseDepthZero-shot=true, # Training samples=320K2024.12 | 0.109 | — | — | — | — | 88.2 | — | — | — | |
| CNM-NetCap=10m2021.12 | 0.1102 | 0.2129 | — | 0.3032 | 0.1482 | 86.88 | — | — | — | |
| MiDaSBackbone=ResNeXt101, #Params=88M, Zero-shot transfer=true, Manually aligned scale and shift=true2023.07 | 0.111 | — | — | — | — | 88.6 | — | — | — | |
| MVDepthNetCap=10m2021.12 | 0.1116 | 0.2087 | — | 0.3143 | 0.15 | 88.04 | — | — | — | |
| IIB2021.12 | 0.1159 | — | — | 0.2807 | — | 0.9079 | — | — | — | |
| MaGNet (D-Net)Cap=5m2021.12 | 0.1177 | 0.1991 | — | 0.2526 | 0.1439 | 85.7 | — | — | — | |
| MaGNet (D-Net)Cap=10m2021.12 | 0.1186 | 0.207 | — | 0.2708 | 0.1461 | 85.46 | — | — | — | |
| MVDepthNetfine-tuned on ScanNet=false, sequence length=3, frame gap=202020.03 | 0.1191 | 0.2096 | 0.091 | 0.3048 | 0.1597 | 86.9 | 95.99 | 98.51 | — | |
| MiDaSZero-shot=true, # Training samples=2M2024.12 | 0.121 | — | — | — | — | 84.6 | — | — | — | |
| VNLSupervision=D2023.12 | 0.123 | — | — | 0.306 | 0.052 | 84.8 | 96.4 | 99.1 | — | |
| DPSNet2021.12 | 0.1258 | — | — | 0.3145 | — | — | — | — | — | |
| GPMVSNetfine-tuned on ScanNet=false, sequence length=3, frame gap=202020.03 | 0.1306 | 0.26 | 0.0944 | 0.3451 | 0.1881 | 84.81 | 94.62 | 97.53 | — | |
| DPSNetfine-tuned on ScanNet=false, sequence length=3, frame gap=202020.03 | 0.137 | 0.2248 | 0.1035 | 0.3468 | 0.1952 | 84.86 | 94.74 | 97.61 | — | |
| FCRNSupervision=D2023.12 | 0.141 | — | — | 0.339 | 0.059 | 81.1 | 95.8 | 99 | — | |
| IndoorDepthSupervision=M2023.12 | 0.153 | — | — | 0.373 | 0.065 | 78.6 | 95 | 98.8 | — | |
| BA-NetTime (s)=0.382022.11 | 0.161 | — | 0.092 | 0.346 | 0.214 | — | — | — | — | |
| BA-NetTime (s)=0.382022.11 | 0.161 | — | 0.092 | 0.346 | 0.214 | — | — | — | — | |
| StructDepthSupervision=M2023.12 | 0.165 | — | — | 0.4 | 0.07 | 75.4 | 93.9 | 98.5 | — | |
| PLNetSupervision=M, frames=52023.12 | 0.168 | — | — | 0.404 | 0.072 | 75 | 94.2 | 98.5 | — | |
| SC-DepthSupervision=M2023.12 | 0.169 | — | — | 0.392 | 0.072 | 74.9 | 93.8 | 98.3 | — | |
| IFMNetSupervision=M2023.12 | 0.17 | — | — | 0.402 | 0.071 | 75.8 | 94 | 98.9 | — | |
| P2NetSupervision=M2023.12 | 0.175 | — | — | 0.42 | 0.074 | 74 | 93.2 | 98.2 | — | |
| PLNetSupervision=M, frames=32023.12 | 0.176 | — | — | 0.414 | 0.074 | 73.5 | 93.9 | 98.5 | — | |
| TrainFlowSupervision=M2023.12 | 0.179 | — | — | 0.415 | 0.076 | 72.6 | 92.7 | 98 | — | |
| Monodepth2Supervision=M2023.12 | 0.2 | — | — | 0.458 | 0.083 | 67.2 | 92.2 | 98.1 | — | |
| MovingIndoorSupervision=M2023.12 | 0.212 | — | — | 0.483 | 0.088 | 65 | 90.5 | 97.6 | — | |
| DeMoNTime (s)=0.692022.11 | 0.231 | — | 0.52 | 0.761 | 0.289 | — | — | — | — | |
| DeMoNTime (s)=0.692022.11 | 0.231 | — | 0.52 | 0.761 | 0.289 | — | — | — | — | |
| DepthProType (Original Task Modality)=Monocular (M), Zero-shot evaluation=true, FPS_GPU=2.8, FPS_CPU=0.22025.12 | — | — | — | — | — | 60.7 | — | — | 59.6 | |
| MASt3RType (Original Task Modality)=Stereo (S), Zero-shot evaluation=true, FPS_GPU=2.7, FPS_CPU=0.72025.12 | — | — | — | — | — | 54.4 | — | — | 60.5 | |
| Metric3Dv2Type (Original Task Modality)=Monocular (M), Zero-shot evaluation=true, FPS_GPU=7.1, FPS_CPU=0.72025.12 | — | — | — | — | — | 82 | — | — | 88.5 | |
| UniDepthV2Type (Original Task Modality)=Monocular (M), Zero-shot evaluation=true, FPS_GPU=13.4, FPS_CPU=1.02025.12 | — | — | — | — | — | 96.2 | — | — | 85.7 | |
| VeloDepthType (Original Task Modality)=Video (V), Zero-shot evaluation=true, FPS_GPU=26.2, FPS_CPU=2.72025.12 | — | — | — | — | — | 91.9 | — | — | 91.4 |