Semantic Scene Completion on SemanticKITTI (val)
40.1mIoU (Mean IoU)SSA-SC
Evaluation results
| Method | Links | |||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| SSA-SCInput priors: Sem.=✓ Oracle, Input priors: Vis.=✓2026.06 | 40.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 59.8 | |
| TALoSoptimization=multi-frame test-time optim.2026.06 | 39.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 56.1 | |
| SCPNetnotes=non-reproducible2026.06 | 37.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 50.2 | |
| JS3C-NetInput priors: Sem.=✓ Oracle, Input priors: Vis.=✓2026.06 | 33.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 57.8 | |
| SemCity-AEInput priors: Sem.=✓ Oracle, Input priors: Vis.=✓2026.06 | 33.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 56.4 | |
| S3CNetnotes=no code2026.06 | 33.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 57.2 | |
| LMSCNet-SSInput priors: Sem.=✓ Oracle, Input priors: Vis.=✓2026.06 | 31 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 57.7 | |
| SSA-SCInput priors: Sem.=✓ WI-TTA, Input priors: Vis.=✓2026.06 | 28.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 59.1 | |
| S3CNetdimension=2D2020.12 | 27.003 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DPS2CNetInput priors: Sem.=✗ [paper], Input priors: Vis.=✗2026.06 | 26.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 60.8 | |
| DiffSSCInput priors: Sem.=✗ [paper], Input priors: Vis.=✗2026.06 | 26.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 60.3 | |
| DPS2CNetCategory=SSC2026.06 | 26.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 60.8 | |
| DiffSSCCategory=SSC2026.06 | 26.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 60.3 | |
| SemCity-AEInput priors: Sem.=✓ WI-TTA, Input priors: Vis.=✓2026.06 | 25.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 56.7 | |
| SemCity refinedetails=GT occupancy leakage2026.06 | 25.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 60.7 | |
| JS3C-NetInput priors: Sem.=✓ WI-TTA, Input priors: Vis.=✓2026.06 | 24.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 56.7 | |
| SSA-SCInput priors: Sem.=✗ [paper], Input priors: Vis.=✗2026.06 | 24.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 58.3 | |
| JS3C-NetInput priors: Sem.=✗ [paper], Input priors: Vis.=✗2026.06 | 24 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 57 | |
| JS3C-NetCategory=SSC2026.06 | 24 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 57 | |
| LMSCNet-SSInput priors: Sem.=✓ WI-TTA, Input priors: Vis.=✓2026.06 | 23 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 57.4 | |
| LODEInput priors: Sem.=✗ [paper], Input priors: Vis.=✗2026.06 | 20.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 51.2 | |
| SalsaNext2020.12 | 17.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| TS3DInput priors: Sem.=✗ [paper], Input priors: Vis.=✗2026.06 | 17.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 50.6 | |
| TS3DCategory=SSC2026.06 | 17.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 50.6 | |
| VOICTimes (s)=0.262, Params (M)=45.42025.12 | 17.42 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CGFormer + PNAMRefinement=PNAM2025.12 | 17.27 | 65.27 | 31.9 | 20.87 | 16 | 22.7 | 34.11 | 23.71 | 569 | 294 | 8.36 | 26.63 | 11.42 | 38.91 | 264 | 437 | 0 | 9.08 | 11.78 | 7.66 | 44.33 | |
| CGFormer + VLGMRefinement=VLGM2025.12 | 17.21 | 65.52 | 32.31 | 20.79 | 13 | 23.27 | 34.33 | 22.33 | 524 | 301 | 7.81 | 26.69 | 10.73 | 39.29 | 270 | 412 | 0 | 8.95 | 11.93 | 7.82 | 43.2 | |
| CGFormer + 3D U-NetRefinement=3D U-Net2025.12 | 17.17 | 65.29 | 32.14 | 20.26 | 13 | 23.11 | 33.99 | 22.39 | 528 | 311 | 8.22 | 26.84 | 11.17 | 38.99 | 265 | 405 | 0 | 8.93 | 11.93 | 7.84 | 43.53 | |
| HTCL-SInput=temporal stereo images (S-T)2024.07 | 17.13 | 63.7 | 32.48 | 23.27 | 0.14 | 24.13 | 34.3 | 20.72 | 3.99 | 2.8 | 11.99 | 26.96 | 8.79 | 37.73 | 2.56 | 2.3 | 0 | 11.22 | 11.49 | 6.95 | 45.51 | |
| LightOccutilization of depth distribution predicted by pre-trained model=true2024.12 | 17.08 | 63 | 33.01 | 22.86 | 1.63 | 25.15 | 30.82 | 15.1 | 4.27 | 6.44 | 13.98 | 25.78 | 9.79 | 36.45 | 6.05 | 2.5 | 0 | 10.59 | 11.85 | 5.22 | 43.83 | |
| LMSCNetCategory=SSC2026.06 | 17 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 55.7 | |
| CGFormerInput=Stereo, Backbone=EfficientNetB72024.05 | 16.87 | 65.51 | 32.31 | 20.82 | 0.16 | 23.52 | 34.32 | 19.44 | 4.61 | 2.71 | 7.67 | 26.93 | 8.83 | 39.54 | 2.38 | 4.08 | 0 | 9.2 | 10.67 | 7.84 | 45.99 | |
| CGFormerRefinement=None2025.12 | 16.87 | 65.51 | 32.31 | 20.82 | 16 | 23.52 | 34.32 | 19.44 | 461 | 271 | 7.67 | 26.93 | 8.83 | 39.54 | 238 | 408 | 0 | 9.2 | 10.67 | 7.84 | 45.99 | |
| DBLIDARNet2020.12 | 16.82 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LMSCNet-SSInput priors: Sem.=✗ [retrained], Input priors: Vis.=✗2026.06 | 16.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 55.6 | |
| SemCity-AEInput priors: Sem.=✗, Input priors: Vis.=✗2026.06 | 16.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 53.9 | |
| LightOccsupervision using dense depth predictions rather than sparse LiDAR depth labels=true2024.12 | 15.05 | 59.72 | 30.34 | 22.5 | 0.16 | 16.99 | 27.97 | 25.25 | 3.55 | 4.27 | 17.42 | 18.71 | 3.94 | 31.94 | 5.14 | 2.08 | 0 | 9.42 | 5.44 | 1.14 | 37.39 | |
| LightOcc2024.12 | 14.93 | 60.08 | 30.58 | 22.8 | 0.11 | 16.04 | 28.1 | 21.88 | 3.88 | 4.45 | 17.85 | 18.77 | 3.74 | 31.88 | 4.98 | 2.38 | 0 | 8.99 | 5.32 | 1.79 | 37.25 | |
| Symphonies2023.06 | 14.89 | 56.37 | 27.58 | 15.28 | 0.95 | 21.64 | 28.68 | 20.44 | 2.54 | 2.82 | 13.89 | 25.72 | 6.6 | 30.87 | 3.52 | 2.24 | 0 | 8.4 | 9.57 | 5.76 | 41.92 | |
| SymphonizeInput=Stereo, Backbone=MaskDINO2024.05 | 14.89 | 56.37 | 27.58 | 15.28 | 0.95 | 21.64 | 28.68 | 20.44 | 2.54 | 2.82 | 13.89 | 25.72 | 6.6 | 30.87 | 3.52 | 2.24 | 0 | 8.4 | 9.57 | 5.76 | 41.92 | |
| SymphonizeTimes (s)=0.319, Params (M)=59.32025.12 | 14.89 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SymphonizeRefinement=None2025.12 | 14.89 | 56.37 | 27.58 | 15.28 | 95 | 21.64 | 28.68 | 20.44 | 254 | 282 | 13.89 | 25.72 | 6.6 | 30.87 | 352 | 224 | 0 | 8.4 | 9.57 | 5.76 | 41.92 | |
| Symphoniesutilization of depth distribution predicted by pre-trained model=true2024.12 | 14.89 | 56.37 | 27.58 | 15.28 | 0.95 | 21.64 | 28.68 | 20.44 | 2.54 | 2.82 | 13.89 | 25.72 | 6.6 | 30.87 | 3.52 | 2.24 | 0 | 8.4 | 9.57 | 5.76 | 41.92 | |
| HDUNet2020.12 | 14.86 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| HASSC-TInput=Stereo-T, Backbone=ResNet502024.05 | 14.74 | 55.3 | 29.6 | 25.9 | 11.3 | 23.1 | 23 | 2.9 | 1.9 | 1.5 | 4.9 | 24.8 | 9.8 | 26.5 | 1.4 | 3 | 0 | 14.3 | 7 | 7.1 | 44.58 | |
| H2GFormer-TInput=Stereo-T, Backbone=ResNet502024.05 | 14.29 | 57 | 29.37 | 21.74 | 0.34 | 20.51 | 28.21 | 6.8 | 0.95 | 0.91 | 9.32 | 27.44 | 7.8 | 36.26 | 1.15 | 0.1 | 0 | 7.98 | 9.88 | 5.81 | 44.69 | |
| OccGen2024.12 | 13.74 | 61.28 | 28.3 | 20.42 | 0.43 | 14.49 | 26.83 | 15.49 | 1.6 | 2.53 | 12.83 | 20.04 | 3.94 | 32.44 | 3.2 | 3.37 | 0 | 6.94 | 4.11 | 2.77 | 36.87 | |
| H2GFormer-SInput=Stereo, Backbone=ResNet502024.05 | 13.73 | 56.08 | 29.12 | 17.83 | 0.45 | 19.74 | 28.21 | 10 | 0.5 | 0.47 | 7.39 | 26.25 | 6.8 | 34.42 | 1.54 | 2.88 | 0 | 7.24 | 7.88 | 4.68 | 44.57 | |
| H2GFormer-SRefinement=None2025.12 | 13.73 | 56.08 | 29.12 | 17.83 | 45 | 19.74 | 28.21 | 10 | 50 | 47 | 7.39 | 26.25 | 6.8 | 34.42 | 154 | 288 | 0 | 7.24 | 7.88 | 4.68 | 44.57 | |
| HASSC-SInput=Stereo, Backbone=ResNet502024.05 | 13.48 | 57.05 | 28.25 | 15.9 | 1.04 | 19.05 | 27.23 | 9.91 | 0.92 | 0.86 | 5.61 | 25.48 | 6.15 | 32.94 | 2.8 | 4.71 | 0 | 6.58 | 7.68 | 4.05 | 44.82 | |
| HASSC-SRefinement=None2025.12 | 13.48 | 57.05 | 28.25 | 15.9 | 104 | 19.05 | 27.23 | 9.91 | 92 | 86 | 5.61 | 25.48 | 6.15 | 32.94 | 280 | 471 | 0 | 6.58 | 7.68 | 4.05 | 44.82 | |
| OccFormerreproduced=true2023.06 | 13.46 | 58.85 | 26.88 | 19.61 | 0.31 | 14.4 | 25.09 | 25.53 | 0.81 | 1.19 | 8.52 | 19.63 | 3.93 | 32.62 | 2.78 | 2.82 | 0 | 5.61 | 4.26 | 2.86 | 36.5 | |
| OccFormer2023.12 | 13.46 | 58.84 | 26.88 | 19.61 | 0.31 | 14.4 | 25.09 | 25.53 | 0.81 | 1.19 | 8.52 | 19.63 | 3.93 | 32.63 | 2.78 | 2.82 | 0 | 5.61 | 4.26 | 2.86 | 36.5 | |
| OccFormerInput=Camera, FLOPs=889G2024.04 | 13.46 | 58.85 | 26.88 | 19.61 | 0.31 | 14.4 | 25.09 | 25.53 | 0.81 | 1.19 | 8.52 | 19.63 | 3.93 | 32.62 | 2.78 | 2.82 | 0 | 5.61 | 4.26 | 2.86 | 36.5 | |
| OccFormerInput=Mono, Backbone=EfficientNetB72024.05 | 13.46 | 58.85 | 26.88 | 19.61 | 0.31 | 14.4 | 25.09 | 25.53 | 0.81 | 1.19 | 8.52 | 19.63 | 3.93 | 32.62 | 2.78 | 2.82 | 0 | 5.61 | 4.26 | 2.86 | 36.5 | |
| OccFormerTimes (s)=0.338, Params (M)=203.42025.12 | 13.46 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| OccFormerRefinement=None2025.12 | 13.46 | 58.85 | 26.88 | 19.61 | 31 | 14.4 | 25.09 | 25.53 | 81 | 119 | 8.52 | 19.63 | 3.93 | 32.62 | 278 | 282 | 0 | 5.61 | 4.26 | 2.86 | 36.5 | |
| OccFormer2024.12 | 13.46 | 58.85 | 26.88 | 19.61 | 0.31 | 14.4 | 25.09 | 25.53 | 0.81 | 1.19 | 8.52 | 19.63 | 3.93 | 32.62 | 2.78 | 2.82 | 0 | 5.61 | 4.26 | 2.86 | 36.5 | |
| Symphonies2023.12 | 13.44 | 55.78 | 26.77 | 14.57 | 0.19 | 18.76 | 27.23 | 15.99 | 1.44 | 2.28 | 9.52 | 24.5 | 4.32 | 28.49 | 3.19 | 8.09 | 0 | 6.18 | 8.99 | 5.39 | 41.44 | |
| VoxFormer-TInput=Stereo-T, Backbone=ResNet502024.05 | 13.35 | 53.57 | 26.52 | 19.69 | 0.42 | 19.54 | 26.54 | 7.26 | 1.28 | 0.56 | 7.81 | 26.1 | 6.1 | 33.06 | 1.93 | 1.97 | 0 | 7.31 | 9.15 | 4.94 | 44.15 | |
| VoxFormer-TInput=temporal stereo images (S-T)2024.07 | 13.35 | 53.57 | 26.52 | 19.69 | 0.42 | 19.54 | 26.54 | 8.1 | 1.28 | 0.56 | 7.81 | 26.1 | 6.1 | 33.06 | 1.93 | 1.97 | 0 | 7.31 | 9.15 | 4.94 | 44.15 | |
| VoxFormerTimes (s)=0.256, Params (M)=57.92025.12 | 13.35 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| DepthSSCInput=Stereo, Backbone=ResNet502024.05 | 13.28 | 55.38 | 27.04 | 18.76 | 0.92 | 19.23 | 25.94 | 6.02 | 0.35 | 1.16 | 7.5 | 26.37 | 4.52 | 30.19 | 2.58 | 6.32 | 0 | 8.46 | 7.42 | 4.09 | 45.84 | |
| DepthSSCRefinement=None2025.12 | 13.28 | 55.38 | 27.04 | 18.76 | 92 | 19.23 | 25.94 | 6.02 | 35 | 116 | 7.5 | 26.37 | 4.52 | 30.19 | 258 | 632 | 0 | 8.46 | 7.42 | 4.09 | 45.84 | |
| SPOT-OccInput=C2026.02 | 13.27 | 59.33 | 28.46 | 21.04 | 1.26 | 15.19 | 26.42 | 16.1 | 1.08 | 1.59 | 11.88 | 18.97 | 303 | 31.57 | 1.82 | 1.16 | 0 | 661 | 4.1 | 2.56 | 36.4 | |
| OctreeOcc2023.12 | 13.12 | 55.13 | 26.74 | 18.68 | 0.65 | 18.69 | 28.07 | 16.43 | 0.64 | 0.71 | 6.03 | 25.26 | 4.89 | 32.47 | 2.25 | 2.57 | 0 | 4.01 | 3.72 | 2.36 | 44.71 | |
| SparseOccInput=Camera, FLOPs=393G2024.04 | 13.12 | 59.59 | 29.68 | 20.44 | 0.47 | 15.41 | 24.03 | 18.07 | 0.78 | 0.89 | 8.94 | 18.89 | 3.46 | 31.06 | 3.68 | 0.62 | 0 | 6.73 | 3.89 | 2.6 | 36.48 | |
| SSv2CRF=false2020.12 | 12.94 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| NDC-SceneSSC Input=RGB2023.09 | 12.7 | 59.2 | 28.24 | 21.42 | 1.67 | 14.94 | 26.26 | 14.75 | 1.67 | 2.37 | 7.73 | 19.09 | 3.51 | 31.04 | 3.6 | 2.74 | 0 | 6.65 | 4.53 | 2.73 | — | |
| NDC-Scene2023.06 | 12.7 | 59.2 | 28.24 | 21.42 | 1.67 | 14.94 | 26.26 | 14.75 | 1.67 | 2.37 | 7.73 | 19.09 | 3.51 | 31.04 | 3.6 | 2.74 | 0 | 6.65 | 4.53 | 2.73 | 37.24 | |
| EditSSCCategory=Editable SSC2026.06 | 12.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 42.1 | |
| IAMSSCInput=Mono, Backbone=ResNet502024.05 | 12.45 | 54.55 | 25.85 | 16.02 | 0.7 | 17.38 | 26.26 | 8.74 | 0.6 | 0.15 | 5.06 | 24.63 | 4.95 | 30.13 | 1.32 | 3.46 | 0.01 | 6.86 | 6.35 | 3.56 | 44.29 | |
| IAMSSCRefinement=None2025.12 | 12.45 | 54.55 | 25.85 | 16.02 | 70 | 17.38 | 26.26 | 8.74 | 60 | 15 | 5.06 | 24.63 | 4.95 | 30.13 | 132 | 346 | 0.01 | 6.86 | 6.35 | 3.56 | 44.29 | |
| VoxFormer2023.06 | 12.35 | 54.76 | 26.35 | 15.5 | 0.7 | 17.65 | 25.79 | 5.63 | 0.59 | 0.51 | 3.77 | 24.39 | 5.08 | 29.96 | 1.78 | 3.32 | 0 | 7.64 | 7.11 | 4.18 | 44.02 | |
| VoxFormer2023.12 | 12.35 | 54.76 | 26.35 | 15.5 | 0.7 | 17.65 | 25.79 | 5.63 | 0.59 | 0.51 | 3.77 | 24.39 | 5.08 | 29.96 | 1.78 | 3.32 | 0 | 7.64 | 7.11 | 4.18 | 44.02 | |
| VoxFormer-SInput=Stereo, Backbone=ResNet502024.05 | 12.35 | 54.76 | 26.35 | 15.5 | 0.7 | 17.65 | 25.79 | 5.63 | 0.59 | 0.51 | 3.77 | 24.39 | 5.08 | 29.96 | 1.78 | 3.32 | 0 | 7.64 | 7.11 | 4.18 | 44.02 | |
| VoxFormer-SInput=single-frame stereo images (S)2024.07 | 12.35 | 54.76 | 26.35 | 15.5 | 0.7 | 17.65 | 25.79 | 7.26 | 0.59 | 0.51 | 3.77 | 24.39 | 5.08 | 29.96 | 1.78 | 3.32 | 0 | 7.64 | 7.11 | 4.18 | 44.02 | |
| VoxFormer-SRefinement=None2025.12 | 12.35 | 54.76 | 26.35 | 15.5 | 70 | 17.65 | 25.79 | 5.63 | 59 | 51 | 3.77 | 24.39 | 5.08 | 29.96 | 178 | 332 | 0 | 7.64 | 7.11 | 4.18 | 44.02 | |
| VoxFormerutilization of depth distribution predicted by pre-trained model=true2024.12 | 12.35 | 54.76 | 26.35 | 15.5 | 0.7 | 17.65 | 25.79 | 5.63 | 0.59 | 0.51 | 3.77 | 24.39 | 5.08 | 29.96 | 1.78 | 3.32 | 0 | 7.64 | 7.11 | 4.18 | 44.02 | |
| SparseOcc†Input=C2026.02 | 12.21 | 59.67 | 28.06 | 13.23 | 0.14 | 15.33 | 25.12 | 9.22 | 0.68 | 1.49 | 8.64 | 19.01 | 351 | 30.93 | 2.29 | 1.21 | 0 | 659 | 4.24 | 2.67 | 35.92 | |
| PointSeg2020.12 | 12.04 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MonoScene + PNAMRefinement=PNAM2025.12 | 11.51 | 56.27 | 26.71 | 14.65 | 145 | 13.48 | 23.11 | 11.38 | 40 | 73 | 3.59 | 17.08 | 2.96 | 29.45 | 295 | 169 | 0 | 6.2 | 3.84 | 2.69 | 36.44 | |
| MonoSceneSSC Input=RGB2023.09 | 11.5 | 57.47 | 27.05 | 15.72 | 0.87 | 14.24 | 23.55 | 7.83 | 0.2 | 0.77 | 3.59 | 18.12 | 2.57 | 30.76 | 1.79 | 1.03 | 0 | 6.39 | 4.11 | 2.48 | — | |
| MonoSceneInput=single-frame monocular images (M)2024.07 | 11.5 | 57.47 | 27.05 | 15.72 | 0.87 | 14.24 | 23.55 | 7.83 | 0.2 | 0.77 | 3.59 | 18.12 | 2.57 | 30.76 | 1.79 | 1.03 | 0 | 6.39 | 4.11 | 2.48 | 37.12 | |
| MonoScene2024.12 | 11.5 | 57.47 | 27.05 | 15.72 | 0.87 | 14.24 | 23.55 | 7.83 | 0.2 | 0.77 | 3.59 | 18.12 | 2.57 | 30.76 | 1.79 | 1.03 | 0 | 6.39 | 4.11 | 2.48 | 37.12 | |
| MonoScene + VLGMRefinement=VLGM2025.12 | 11.49 | 56.48 | 26.64 | 14.35 | 142 | 13.55 | 22.76 | 12.45 | 44 | 71 | 3.12 | 16.44 | 2.97 | 29.5 | 304 | 164 | 0 | 6.28 | 3.85 | 2.65 | 35.62 | |
| MonoScene + 3D U-NetRefinement=3D U-Net2025.12 | 11.47 | 56.24 | 26.63 | 14.95 | 142 | 13.11 | 23.46 | 10.95 | 41 | 87 | 3.69 | 16.75 | 2.73 | 29.57 | 298 | 164 | 0 | 6.19 | 3.77 | 2.62 | 35.7 | |
| TPVFormerreproduced=true2023.06 | 11.36 | 56.5 | 25.87 | 20.6 | 0.85 | 13.88 | 23.81 | 8.08 | 0.36 | 0.05 | 4.35 | 16.92 | 2.26 | 30.38 | 0.51 | 0.89 | 0 | 5.94 | 3.14 | 1.52 | 35.61 | |
| TPVFormer2023.12 | 11.36 | 56.5 | 25.87 | 20.6 | 0.85 | 13.88 | 23.81 | 8.08 | 0.36 | 0.05 | 4.35 | 16.92 | 2.26 | 30.38 | 0.51 | 0.89 | 0 | 5.94 | 3.14 | 1.52 | 35.61 | |
| TPVFormerInput=Camera, FLOPs=946G2024.04 | 11.36 | 56.5 | 25.87 | 20.6 | 0.85 | 13.88 | 23.81 | 8.08 | 0.36 | 0.05 | 4.35 | 16.92 | 2.26 | 30.38 | 0.51 | 0.89 | 0 | 5.94 | 3.14 | 1.52 | 35.61 | |
| TPVFormerInput=Mono, Backbone=EfficientNetB72024.05 | 11.36 | 56.5 | 25.87 | 20.6 | 0.85 | 13.88 | 23.81 | 8.08 | 0.36 | 0.05 | 4.35 | 16.92 | 2.26 | 30.38 | 0.51 | 0.89 | 0 | 5.94 | 3.14 | 1.52 | 35.61 | |
| TPVFormerInput=C2026.02 | 11.36 | 56.5 | 25.87 | 20.6 | 0.85 | 13.88 | 23.81 | 8.08 | 0.36 | 0.05 | 4.35 | 16.92 | 226 | 30.38 | 0.51 | 0.89 | 0 | 594 | 3.14 | 1.52 | 35.61 | |
| TPVFormerRefinement=None2025.12 | 11.36 | 56.5 | 25.87 | 20.6 | 85 | 13.88 | 23.81 | 8.08 | 36 | 5 | 4.35 | 16.92 | 2.26 | 30.38 | 51 | 89 | 0 | 5.94 | 3.14 | 1.52 | 35.61 | |
| TPVFormer2024.12 | 11.36 | 56.5 | 25.87 | 20.6 | 0.85 | 13.88 | 23.81 | 8.08 | 0.36 | 0.05 | 4.35 | 16.92 | 2.26 | 30.38 | 0.51 | 0.89 | 0 | 5.94 | 3.14 | 1.52 | 35.61 | |
| MonoScenereproduced=true2023.06 | 11.08 | 56.52 | 26.72 | 14.27 | 0.46 | 14.09 | 23.26 | 6.98 | 0.61 | 0.45 | 1.48 | 17.89 | 2.81 | 29.64 | 1.86 | 1.2 | 0 | 5.84 | 4.14 | 2.25 | 36.86 | |
| MonoScene2023.12 | 11.08 | 56.52 | 26.72 | 14.27 | 0.46 | 14.09 | 23.26 | 6.98 | 0.61 | 0.45 | 1.48 | 17.89 | 2.81 | 29.64 | 1.86 | 1.2 | 0 | 5.84 | 4.14 | 2.25 | 36.86 | |
| MonoSceneInput=Camera2024.04 | 11.08 | 56.52 | 26.72 | 14.27 | 0.46 | 14.09 | 23.26 | 6.98 | 0.61 | 0.45 | 1.48 | 17.89 | 2.81 | 29.64 | 1.86 | 1.2 | 0 | 5.84 | 4.14 | 2.25 | 36.86 | |
| MonoSceneInput=Mono, Backbone=EfficientNetB72024.05 | 11.08 | 56.52 | 26.72 | 14.27 | 0.46 | 14.09 | 23.26 | 6.98 | 0.61 | 0.45 | 1.48 | 17.89 | 2.81 | 29.64 | 1.86 | 1.2 | 0 | 5.84 | 4.14 | 2.25 | 36.86 | |
| MonoSceneInput=C2026.02 | 11.08 | 56.52 | 26.72 | 14.27 | 0.46 | 14.09 | 23.26 | 6.98 | 0.61 | 0.45 | 1.48 | 17.89 | 281 | 29.64 | 1.86 | 1.2 | 0 | 584 | 4.14 | 2.25 | 36.86 | |
| MonoSceneTimes (s)=0.274, Params (M)=132.42025.12 | 11.08 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — |