Monocular Depth Estimation on KITTI Improved GT (Eigen)
0.059AbsRelSCIPaD
Evaluation results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| SCIPaDtraining=M, Resolution=192 x 6402024.07 | 0.059 | 0.287 | 2.871 | 0.095 | 96.4 | — | — | — | 99.3 | 99.8 | — | |
| Jasminevariant=main2025.03 | 0.061 | 0.255 | 2.765 | 0.092 | 96.3 | 99.5 | 0.999 | — | — | — | — | |
| SQLdepthtraining=M, Resolution=192 x 6402024.07 | 0.061 | 0.317 | 3.055 | 0.1 | 95.7 | — | — | — | 99.2 | 99.7 | — | |
| Flex-X-LargeType=Self-supervised, Params=32M, GFLOPs=25, Resolution=640×1922026.07 | 0.063 | — | — | — | — | — | — | 95.2 | — | — | — | |
| Jasmine*variant=*2025.03 | 0.064 | 0.294 | 2.982 | 0.097 | 95.7 | 99.4 | 0.998 | — | — | — | — | |
| MonoViT2025.03 | 0.068 | 0.314 | 3.125 | 0.105 | 94.8 | 99.2 | 0.998 | — | — | — | — | |
| DynamicDepthtraining=M, Resolution=192 x 6402024.07 | 0.068 | 0.362 | 3.454 | 0.111 | 94.3 | — | — | — | 99.1 | 99.8 | — | |
| TriDepthtraining=M, Resolution=192 x 6402024.07 | 0.068 | 0.359 | 3.341 | 0.11 | 94.4 | — | — | — | 98.9 | 99.7 | — | |
| SA4Depth + Hybrid-depthScaling=Median-scaled (done w/ GT), Training=M, Resolution=640 x 1922026.05 | 0.068 | 0.317 | 3.103 | 0.105 | — | — | — | 95.2 | 99.2 | 99.8 | 2.71 | |
| ManyDepthtraining=M, Resolution=192 x 6402024.07 | 0.07 | 0.399 | 3.455 | 0.113 | 94.1 | — | — | — | 98.9 | 99.7 | — | |
| DA2 (ViT-L)Type=Zero-Shot, Params=335M, GFLOPs=1947, Resolution=1722×5182026.07 | 0.07 | — | — | — | — | — | — | 95.6 | — | — | — | |
| MonoViFi2025.03 | 0.071 | 0.338 | 3.539 | 0.113 | 93.7 | 99 | 0.998 | — | — | — | — | |
| SA4Depth + MonoViTScaling=Median-scaled (done w/ GT), Training=M, Resolution=640 x 1922026.05 | 0.071 | 0.38 | 3.341 | 0.111 | — | — | — | 94.5 | 98.9 | 99.7 | 0.16 | |
| Mono-ViFIScaling=Median-scaled (done w/ GT), Training=M+VI, Resolution=640 x 192, Post-processing=w/o2026.05 | 0.072 | 0.342 | 3.254 | 0.111 | — | — | — | 94.1 | 99.1 | 99.8 | 3.32 | |
| Hybrid-depthScaling=Median-scaled (done w/ GT), Training=M, Resolution=640 x 1922026.05 | 0.072 | 0.334 | 3.264 | 0.11 | — | — | — | 94.4 | 99.1 | 99.8 | 3.56 | |
| Depth Hints Resnet50Post-Processing=true, Data=SSGM, Resolution=320 x 10242021.06 | 0.074 | 0.363 | 3.198 | 0.114 | — | — | — | 93.6 | 98.9 | 99.7 | — | |
| Wavelet Monodepth Resnet50Post-Processing=true, Data=SSGM, Resolution=320 x 10242021.06 | 0.074 | 0.357 | 3.17 | 0.114 | — | — | — | 93.6 | 98.9 | 99.7 | — | |
| DualRefineScaling=Median-scaled (done w/ GT), Training=M, Resolution=640 x 192, Post-processing=w/o2026.05 | 0.075 | 0.379 | 3.49 | 0.117 | — | — | — | 93.6 | 98.9 | 99.7 | 3.39 | |
| MonoViTScaling=Median-scaled (done w/ GT), Training=M, Resolution=640 x 1922026.05 | 0.075 | 0.389 | 3.419 | 0.115 | — | — | — | 93.8 | 98.9 | 99.7 | 3.73 | |
| Wavelet Monodepth Resnet50Post-Processing=true, Data=S, Resolution=320 x 10242021.06 | 0.076 | 0.413 | 3.434 | 0.126 | — | — | — | 92.6 | 98.4 | 99.5 | — | |
| Depth HintsPost-Processing=true, Data=SSGM, Resolution=320 x 10242021.06 | 0.077 | 0.404 | 3.345 | 0.119 | — | — | — | 93 | 98.8 | 99.7 | — | |
| SA4Depth + MonoViTScaling=scale-aware (Learned metric), Training=M+v, Resolution=640 x 1922026.05 | 0.077 | 0.404 | 3.472 | 0.119 | — | — | — | 93.2 | 98.8 | 99.7 | 0.05 | |
| DA2 (ViT-S)Type=Zero-Shot, Params=25M, GFLOPs=137, Resolution=1722×5182026.07 | 0.077 | — | — | — | — | — | — | 94.4 | — | — | — | |
| PackNet-SfMtraining=M, Resolution=192 x 6402024.07 | 0.078 | 0.42 | 3.485 | 0.121 | 93.1 | — | — | — | 98.6 | 99.6 | — | |
| Wavelet Monodepth Resnet18Post-Processing=true, Data=SSGM, Resolution=320 x 10242021.06 | 0.078 | 0.397 | 3.316 | 0.121 | — | — | — | 92.8 | 98.7 | 99.7 | — | |
| PackNet-SfMScaling=Median-scaled (done w/ GT), Training=M, Resolution=640 x 192, Post-processing=w/o2026.05 | 0.078 | 0.42 | 3.485 | 0.121 | — | — | — | 93.1 | 98.6 | 99.6 | 0.75 | |
| Monodepth2 Resnet18Post-Processing=true, Data=S, Resolution=192 x 6402021.06 | 0.079 | 0.512 | 3.721 | 0.131 | — | — | — | 92.4 | 98.2 | 99.4 | — | |
| Monodepth2training=MS, Resolution=192 x 6402024.07 | 0.08 | 0.466 | 3.681 | 0.127 | 92.6 | — | — | — | 98.5 | 99.5 | — | |
| Wavelet Monodepth Resnet18Post-Processing=true, Data=S, Resolution=320 x 10242021.06 | 0.08 | 0.443 | 3.544 | 0.13 | — | — | — | 91.9 | 98.3 | 99.5 | — | |
| SA4Depth + Monodepth2-50Scaling=Median-scaled (done w/ GT), Training=M, Backbone=ResNet-50, Resolution=640 x 1922026.05 | 0.08 | 0.465 | 3.707 | 0.125 | — | — | — | 92.8 | 98.6 | 99.6 | 1.9 | |
| Lotus2025.03 | 0.081 | — | — | — | 93.1 | 98.7 | — | — | — | — | — | |
| Wavelet Monodepth Resnet50Post-Processing=true, Data=S, Resolution=192 x 6402021.06 | 0.081 | 0.477 | 3.658 | 0.133 | — | — | — | 92 | 98.1 | 99.4 | — | |
| Depth Hints Resnet50Post-Processing=true, Data=SSGM, Resolution=192 x 6402021.06 | 0.081 | 0.432 | 3.51 | 0.124 | — | — | — | 92.4 | 98.5 | 99.6 | — | |
| Wavelet Monodepth Resnet50Post-Processing=true, Data=SSGM, Resolution=192 x 6402021.06 | 0.081 | 0.449 | 3.509 | 0.125 | — | — | — | 92.3 | 98.6 | 99.6 | — | |
| Monodepth2 Resnet18Post-Processing=true, Data=S, Resolution=320 x 10242021.06 | 0.082 | 0.497 | 3.637 | 0.132 | — | — | — | 92.4 | 98.2 | 99.4 | — | |
| Wavelet Monodepth Resnet18Post-Processing=true, Data=SSGM, Resolution=192 x 6402021.06 | 0.083 | 0.476 | 3.635 | 0.129 | — | — | — | 92 | 98.3 | 99.5 | — | |
| Wavelet Monodepth Resnet18Post-Processing=true, Data=S, Resolution=192 x 6402021.06 | 0.084 | 0.523 | 3.807 | 0.137 | — | — | — | 91.4 | 98 | 99.4 | — | |
| Monodepth2 Resnet18Post-Processing=true, Data=MS, Resolution=192 x 6402021.06 | 0.084 | 0.494 | 3.739 | 0.132 | — | — | — | 91.8 | 98.3 | 99.5 | — | |
| Depth HintsPost-Processing=true, Data=SSGM, Resolution=192 x 6402021.06 | 0.085 | 0.487 | 3.67 | 0.131 | — | — | — | 91.7 | 98.3 | 99.6 | — | |
| Wavelet Monodepth Resnet18Post-Processing=true, Data=MS, Resolution=192 x 6402021.06 | 0.085 | 0.497 | 3.804 | 0.134 | — | — | — | 91.2 | 98.2 | 99.5 | — | |
| Monodepth2-50Scaling=Median-scaled (done w/ GT), Training=M, Backbone=ResNet-50, Resolution=640 x 1922026.05 | 0.085 | 0.468 | 3.674 | 0.128 | — | — | — | 92.1 | 98.5 | 99.6 | 2.65 | |
| SA4Depth + Monodepth2-50Scaling=scale-aware (Learned metric), Training=M+v, Backbone=ResNet-50, Resolution=640 x 1922026.05 | 0.085 | 0.476 | 3.782 | 0.133 | — | — | — | 91.4 | 98.3 | 99.5 | 0.06 | |
| Wavelet Monodepth Resnet18Post-Processing=true, Data=MS + SSGM, Resolution=192 x 6402021.06 | 0.086 | 0.497 | 3.699 | 0.131 | — | — | — | 91.4 | 98.3 | 99.6 | — | |
| Depth HintsPost-Processing=true, Data=MS + SSGM, Resolution=192 x 6402021.06 | 0.087 | 0.526 | 3.776 | 0.133 | — | — | — | 91.5 | 98.2 | 99.5 | — | |
| SA4Depth + Monodepth2-18Scaling=Median-scaled (done w/ GT), Training=M, Backbone=ResNet-18, Resolution=640 x 1922026.05 | 0.087 | 0.524 | 3.91 | 0.134 | — | — | — | 91.8 | 98.2 | 99.5 | 0.09 | |
| SA4Depth + Monodepth2-18Scaling=scale-aware (Learned metric), Training=M+v, Backbone=ResNet-18, Resolution=640 x 1922026.05 | 0.088 | 0.556 | 3.938 | 0.138 | — | — | — | 91 | 98 | 99.4 | 0.06 | |
| MonoViTScaling=scale-aware (Learned metric), Training=M+v, Resolution=640 x 1922026.05 | 0.089 | 0.466 | 3.59 | 0.129 | — | — | — | 91.3 | 98.5 | 99.6 | 0.07 | |
| JasmineNotes=Mono, Data=KH(68K+0), Alignment=Alternative alignment discussed in Sec. 4.32025.03 | 0.09 | 0.581 | 3.944 | 0.161 | 91.9 | 97.2 | 98.6 | — | — | — | — | |
| Monodepth2-18Scaling=Median-scaled (done w/ GT), Training=M, Backbone=ResNet-18, Resolution=640 x 1922026.05 | 0.09 | 0.545 | 3.942 | 0.137 | — | — | — | 91.4 | 98.3 | 99.5 | 2.79 | |
| Jaeho et al*Venue=CVPR2024, Notes=Mono, Data=K(40K+0)2025.03 | 0.091 | 0.604 | 4.066 | 0.164 | 91.3 | 97 | 98.6 | — | — | — | — | |
| RPrDepthVenue=ECCV2024, Notes=Mono, Data=K(40K+0)2025.03 | 0.091 | 0.612 | 4.098 | 0.162 | 91 | 97.1 | 98.6 | — | — | — | — | |
| FUMETScaling=scale-aware (Learned metric), Training=M+SI, Resolution=640 x 1922026.05 | 0.091 | 0.533 | 3.892 | 0.136 | — | — | — | 90.6 | 98.3 | 99.5 | 0.07 | |
| Monodepth2-18Scaling=scale-aware (Learned metric), Training=M+v, Backbone=ResNet-18, Resolution=640 x 1922026.05 | 0.092 | 0.549 | 3.995 | 0.142 | — | — | — | 90.2 | 98 | 99.4 | 0.07 | |
| DA2 (ViT-L)Type=Zero-Shot, Params=335M, GFLOPs=276, Resolution=644×1962026.07 | 0.092 | — | — | — | — | — | — | 91.5 | — | — | — | |
| Mono-ViFIVenue=ECCV2024, Notes=Mono, Data=K(40K+0)2025.03 | 0.093 | 0.589 | 4.072 | 0.168 | 90.9 | 96.9 | 98.5 | — | — | — | — | |
| VADepthScaling=scale-aware (Learned metric), Training=M+camH, Resolution=640 x 1922026.05 | 0.093 | 0.493 | 3.704 | 0.132 | — | — | — | 91.3 | 98.4 | 99.6 | 0.07 | |
| Monodepth2-50Scaling=scale-aware (Learned metric), Training=M+v, Backbone=ResNet-50, Resolution=640 x 1922026.05 | 0.093 | 0.559 | 3.984 | 0.14 | — | — | — | 90.1 | 98.1 | 99.5 | 0.07 | |
| DaCCNVenue=ICCV2023, Notes=Mono, Data=K(40K+0)2025.03 | 0.094 | 0.624 | 4.145 | 0.169 | 90.9 | 97 | 98.5 | — | — | — | — | |
| E2E-FT2025.03 | 0.096 | — | — | — | 92.1 | 98 | — | — | — | — | — | |
| MonoViTVenue=3DV 2022, Notes=Mono, Data=K(40K+0)2025.03 | 0.096 | 0.714 | 4.292 | 0.172 | 90.8 | 96.8 | 98.4 | — | — | — | — | |
| SD-SSMDEVenue=CVPR2022, Notes=Mono, Data=K(40K+0)2025.03 | 0.098 | 0.674 | 4.187 | 0.17 | 90.2 | 96.8 | 98.5 | — | — | — | — | |
| Marigold2025.03 | 0.099 | — | — | — | 91.6 | 98.7 | — | — | — | — | — | |
| DepthSegNetVenue=ECCV2022, Notes=Mono, Data=K(40K+0)2025.03 | 0.099 | 0.624 | 4.165 | 0.171 | 90.2 | 96.9 | 98.5 | — | — | — | — | |
| DevNetVenue=ECCV2022, Notes=Mono, Data=K(40K+0)2025.03 | 0.1 | 0.699 | 4.412 | 0.174 | 89.3 | 96.6 | 98.5 | — | — | — | — | |
| Jasmine*Notes=Mono, Data=KH(68K+0), Alignment=Standard alignment discussed in Sec. 4.32025.03 | 0.102 | 0.54 | 3.728 | 0.162 | 90.7 | 97.3 | 98.7 | — | — | — | — | |
| LiteMonoVenue=CVPR2023, Notes=Mono, Data=K(40K+0)2025.03 | 0.102 | 0.746 | 4.444 | 0.179 | 89.6 | 96.5 | 98.3 | — | — | — | — | |
| HR-DepthVenue=AAAI2021, Notes=Mono, Data=K(40K+0)2025.03 | 0.106 | 0.755 | 4.472 | 0.181 | 89.2 | 96.6 | 98.4 | — | — | — | — | |
| R-MSFM6Venue=ICCV2021, Notes=Mono, Data=K(40K+0)2025.03 | 0.108 | 0.748 | 4.47 | 0.185 | 88.9 | 96.3 | 98.2 | — | — | — | — | |
| monodepthSupervised=false, Training Dataset=CS + K, Backbone=ResNet, Post-processing=true, Evaluation cap=50m2016.09 | 0.108 | 0.657 | 3.729 | 0.194 | 87.3 | 95.4 | 97.9 | — | — | — | — | |
| Wang et al.Supervision=M2023.04 | 0.109 | 0.779 | 4.641 | 0.186 | 88.3 | 96.2 | 98.2 | — | — | — | — | |
| LotusVenue=ICLR2025, Notes=ZeroShot, Data=Syn(59K+59K)2025.03 | 0.11 | 0.611 | 3.807 | 0.175 | 89.2 | 97 | 98.6 | — | — | — | — | |
| DA2 (ViT-S)Type=Zero-Shot, Params=25M, GFLOPs=19, Resolution=644×1962026.07 | 0.11 | — | — | — | — | — | — | 88.1 | — | — | — | |
| PackNet-SfMSupervision=M+v, Backbone=ResNet-182023.04 | 0.111 | 0.829 | 4.788 | 0.199 | 86.4 | 95.4 | 98 | — | — | — | — | |
| E2E FTVenue=WACV2025, Notes=ZeroShot, Data=Syn(74K+74K)2025.03 | 0.112 | 0.649 | 4.099 | 0.18 | 89 | 96.9 | 98.5 | — | — | — | — | |
| G2SSupervision=M+v, Backbone=ResNet-502023.04 | 0.112 | 0.894 | 4.852 | 0.192 | 87.7 | 95.8 | 98.1 | — | — | — | — | |
| VA-DepthSupervision=M+h2023.04 | 0.112 | 0.864 | 4.804 | 0.19 | 87.8 | 95.9 | 98.2 | — | — | — | — | |
| monodepthSupervised=false, Training Dataset=CS + K, Post-processing=true, Evaluation cap=50m2016.09 | 0.112 | 0.68 | 3.81 | 0.198 | 86.6 | 95.3 | 97.9 | — | — | — | — | |
| DNetSupervision=M+h2023.04 | 0.113 | 0.864 | 4.812 | 0.191 | 87.7 | 96 | 98.1 | — | — | — | — | |
| pRGBD-RefinedSupervision=M+SLAM2023.04 | 0.113 | 0.793 | 4.655 | 0.188 | 87.4 | 96 | 98.3 | — | — | — | — | |
| Kuznietsov et al.Supervision=M+D2023.04 | 0.113 | 0.741 | 4.621 | 0.189 | 86.2 | 96 | 98.6 | — | — | — | — | |
| TrainFlowSupervision=M+SfM2023.04 | 0.113 | 0.704 | 4.581 | 0.184 | 87.1 | 96.1 | 98.4 | — | — | — | — | |
| TC-DepthSupervision=M, config=only Lgeo2023.04 | 0.113 | 0.904 | 4.773 | 0.193 | 87.7 | 95.9 | 98 | — | — | — | — | |
| OursSupervision=M2023.04 | 0.113 | 0.851 | 4.809 | 0.191 | 87.7 | 95.9 | 98.1 | — | — | — | — | |
| monodepthSupervised=false, Training Dataset=CS + K, Backbone=ResNet, Post-processing=true2016.09 | 0.114 | 0.898 | 4.935 | 0.206 | 86.1 | 94.9 | 97.6 | — | — | — | — | |
| Monodepth2Venue=ICCV2019, Notes=Mono, Data=K(40K+0)2025.03 | 0.115 | 0.882 | 4.701 | 0.19 | 87.9 | 96.1 | 98.2 | — | — | — | — | |
| Monodepth2Supervision=M2023.04 | 0.115 | 0.903 | 4.863 | 0.193 | 87.7 | 95.9 | 98.1 | — | — | — | — | |
| monodepthSupervised=false, Training Dataset=CS + K, Evaluation cap=50m2016.09 | 0.117 | 0.762 | 3.972 | 0.206 | 86 | 94.8 | 97.6 | — | — | — | — | |
| Sparse-to-ContSupervision=M+D2023.04 | 0.118 | 0.63 | 4.52 | 0.209 | 89.8 | 96.6 | 98.5 | — | — | — | — | |
| monodepthSupervised=false, Training Dataset=CS + K, Post-processing=true2016.09 | 0.118 | 0.923 | 5.015 | 0.21 | 85.4 | 94.7 | 97.6 | — | — | — | — | |
| SC-SfMLearnerSupervision=M2023.04 | 0.119 | 0.857 | 4.95 | 0.197 | 86.3 | 95.7 | 98.1 | — | — | — | — | |
| MarigoldVenue=CVPR2024, Notes=ZeroShot, Data=Syn(74K+74K)2025.03 | 0.12 | 0.672 | 4.033 | 0.184 | 87.4 | 96.8 | 98.5 | — | — | — | — | |
| monodepthSupervised=false, Training Dataset=CS + K2016.09 | 0.124 | 1.076 | 5.311 | 0.219 | 84.7 | 94.2 | 97.3 | — | — | — | — | |
| Godard et al.Training Dataset=Cityscapes + KITTI, Pose Supervision=true2017.04 | 0.124 | 1.076 | 5.311 | 0.219 | — | — | — | 84.7 | 94.2 | 97.3 | — | |
| Li et al.Supervision=M2023.04 | 0.13 | 0.95 | 5.138 | 0.209 | 84.3 | 94.8 | 97.8 | — | — | — | — | |
| monodepthSupervised=false, Training Dataset=CS + K, Backbone=ResNet, Post-processing=true, Crop Strategy=uncropped2016.09 | 0.13 | 1.197 | 5.222 | 0.226 | 84.3 | 94 | 97.1 | — | — | — | — | |
| monodepthSupervised=false, Training Dataset=CS + K, Post-processing=true, Crop Strategy=uncropped2016.09 | 0.134 | 1.261 | 5.336 | 0.23 | 83.5 | 93.8 | 97.1 | — | — | — | — | |
| Sheng et al.Supervision=M2023.04 | 0.139 | 1.021 | 5.418 | 0.209 | 80.3 | 93.7 | 97.6 | — | — | — | — | |
| monodepthSupervised=false, Training Dataset=K, Evaluation cap=50m2016.09 | 0.14 | 0.976 | 4.471 | 0.232 | 81.8 | 93.1 | 96.9 | — | — | — | — | |
| monodepthSupervised=false, Training Dataset=K2016.09 | 0.148 | 1.344 | 5.927 | 0.247 | 80.3 | 92.2 | 96.4 | — | — | — | — | |
| Godard et al.Training Dataset=KITTI, Pose Supervision=true2017.04 | 0.148 | 1.344 | 5.927 | 0.247 | — | — | — | 80.3 | 92.2 | 96.4 | — |