3D semantic occupancy prediction on Occ3D
51.8RayIoUALOcc-GF
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| ALOcc-GFBackbone=Swin-Base, Input Size=512 x 14082025.04 | 51.8 | — | — | — | 47.4 | 79.9 | 1.8 | 8,312 | — | |
| ALOccBackbone=Swin-Base, Input Size=512 x 14082025.04 | 50.6 | — | — | — | 46.1 | 78.1 | 1.5 | 24,072 | — | |
| HiPRInput=C+L, Backbone=R502026.05 | 50 | 45.9 | 50.5 | 53.4 | — | — | — | — | — | |
| BEVDetOcc-GFBackbone=Swin-Base, Input Size=512 x 14082025.04 | 48.5 | — | — | — | 42.9 | 79.5 | 2.2 | 4,492 | — | |
| DAOccInput=C+L, Backbone=R502026.05 | 48.4 | 44.5 | 48.9 | 51.9 | — | — | — | — | — | |
| HiPR-miniInput=C+L, Backbone=R502026.05 | 48.4 | 44.2 | 49 | 52 | — | — | — | — | — | |
| BEVDetOcc-SFBackbone=Swin-Base, Input Size=512 x 14082025.04 | 47.1 | — | — | — | 41.2 | 76.8 | 2.2 | 13,682 | — | |
| ALOcc-GFBackbone=ResNet-50, Input Size=256 x 7042025.04 | 46.5 | — | — | — | 40.2 | 77.4 | 6.2 | 4,347 | — | |
| COTRBackbone=Swin-Base, Input Size=512 x 14082025.04 | 46.2 | — | — | — | 41.3 | 74.9 | — | — | — | |
| ALOccBackbone=ResNet-50, Input Size=256 x 7042025.04 | 45.5 | — | — | — | 39.3 | 75.3 | 6 | 10,793 | — | |
| COTRBackbone=ResNet-50, Input Size=256 x 7042025.04 | 44.5 | — | — | — | 38.6 | 75 | 0.5 | 8,063 | — | |
| ALOcc-GFBackbone=ResNet-50, Input Size=256×7042025.04 | 44.1 | 38.2 | 45 | 49.2 | — | — | — | — | — | |
| ALOccBackbone=ResNet-50, Input Size=256×7042025.04 | 43.7 | 37.8 | 44.7 | 48.8 | — | — | — | — | — | |
| BEVDetOcc-GFBackbone=ResNet-50, Input Size=256 x 7042025.04 | 43.6 | — | — | — | 36 | 77.8 | 7 | 3,017 | — | |
| ALOcc-2DInput=C, Backbone=R502026.05 | 43 | 37.1 | 43.8 | 48.2 | — | — | — | — | — | |
| SurroundSDFBackbone=ResNet-50, Input Size=256 x 7042025.04 | 42.4 | — | — | — | 36.2 | — | — | — | — | |
| PanoOccBackbone=ResNet-101, Input Size=640 x 16002025.04 | 42.1 | — | — | — | 37.3 | — | — | — | — | |
| STCOccInput=C, Backbone=R502026.05 | 42.1 | 36.9 | 42.8 | 46.7 | — | — | — | — | — | |
| ALOcc-mini-GFBackbone=ResNet-50, Input Size=256 x 7042025.04 | 42 | — | — | — | 35.7 | 71.9 | 23.5 | 1,179 | — | |
| ViewFormerBackbone=ResNet-50, Input Size=256 x 7042025.04 | 41.9 | — | — | — | 35 | 70.2 | — | 2,875 | — | |
| BEVDetOcc-SFBackbone=ResNet-50, Input Size=256 x 7042025.04 | 41.9 | — | — | — | 34.4 | 75.1 | 6.5 | 10,717 | — | |
| FB-Occ-GFBackbone=ResNet-50, Input Size=256 x 7042025.04 | 41.7 | — | — | — | 35.8 | 73.2 | 10.3 | 2,879 | — | |
| ALOcc-miniBackbone=ResNet-50, Input Size=256 x 7042025.04 | 41.4 | — | — | — | 35.4 | 70 | 30.5 | 1,605 | — | |
| OPUSBackbone=ResNet-50, Input Size=256×7042025.04 | 41.2 | 34.7 | 42.1 | 46.7 | — | — | — | — | — | |
| OSPBackbone=ResNet-101, Input Size=900 x 16002025.04 | 41.2 | — | — | — | 37 | — | — | 2,671 | — | |
| OPUSInput=C, Backbone=R502026.05 | 41.2 | 34.7 | 42.1 | 46.7 | — | — | — | — | — | |
| SurroundOccBackbone=InternImage-Base, Input Size=900 x 16002025.04 | 40.7 | — | — | — | — | — | — | — | — | |
| FB-Occ-GFBackbone=ResNet-50, Input Size=256×7042025.04 | 40.6 | 35 | 41.5 | 45.3 | — | — | — | — | — | |
| FB-OccBackbone=ResNet-50, Input Size=256 x 7042025.04 | 39.8 | — | — | — | 34.2 | 69.9 | — | 4,099 | — | |
| UniOCCBackbone=ResNet-50, Input Size=256 x 7042025.04 | 39.7 | — | — | — | — | — | — | — | — | |
| ALOcc-2D-miniInput=C, Backbone=R502026.05 | 39.3 | 32.9 | 40.1 | 44.8 | — | — | — | — | — | |
| BEVFormerBackbone=ResNet-101, Input Size=900 x 16002025.04 | 39.2 | — | — | — | 37.2 | — | 4.4 | 4,667 | — | |
| FB-OccBackbone=ResNet-50, Input Size=256×7042025.04 | 39 | 33 | 40 | 44 | — | — | — | — | — | |
| FB-OccInput=C, Backbone=R502026.05 | 39 | 33 | 40 | 44 | — | — | — | — | — | |
| Panoptic-FlashOccBackbone=ResNet-50, Input Size=256×7042025.04 | 38.5 | 32.8 | 39.3 | 43.4 | — | — | — | — | — | |
| FlashOccInput=C, Backbone=R502026.05 | 38.5 | 32.8 | 39.3 | 43.4 | — | — | — | — | — | |
| BEVDetOcc-GFBackbone=ResNet-50, Input Size=256×7042025.04 | 36.6 | 32.6 | 37.3 | 39.9 | — | — | — | — | — | |
| OPUSBackbone=ResNet-50, Input Size=256 x 7042025.04 | 36.2 | — | — | — | 33.3 | 54 | 8.2 | 2,671 | — | |
| SparseOccBackbone=ResNet-50, Input Size=256×7042025.04 | 36.1 | 30.2 | 36.8 | 41.2 | — | — | — | — | — | |
| SparseOccInput=C, Backbone=R502026.05 | 36.1 | 30.2 | 36.8 | 41.2 | — | — | — | — | — | |
| BEVDetOcc-SOLOBackbone=ResNet-50, Input Size=256×7042025.04 | 35.2 | 31.2 | 35.9 | 38.4 | — | — | — | — | — | |
| FastOccBackbone=ResNet-101, Input Size=640 x 16002025.04 | 34.5 | — | — | — | 39.2 | — | — | — | — | |
| BEVFormerInput=C, Backbone=R1012026.05 | 32.4 | 26.1 | 32.9 | 38 | — | — | — | — | — | |
| FlashOCCBackbone=ResNet-50, Input Size=256 x 7042025.04 | 32 | — | — | — | 24.7 | 65.3 | 29.6 | 1,205 | — | |
| RenderOccInput=C, Backbone=Swin-B2026.05 | 24.4 | 13.4 | 19.6 | 25.5 | — | — | — | — | — | |
| RenderOccBackbone=Swin-Base, Input Size=512×14082025.04 | 19.5 | 13.4 | 19.6 | 25.5 | — | — | — | — | — | |
| SQOccRep.=Quad2025.11 | 17.3 | 11.9 | 17.3 | 22.56 | — | — | — | — | — | |
| GaussianFlowOccRep.=Gauss2025.11 | 16.2 | 11.2 | 16.2 | 21.2 | — | — | — | — | — | |
| TT-OccRep.=Gauss2025.11 | 13.4 | 10 | 12.9 | 15.7 | — | — | — | — | — | |
| GaussianOccRep.=Voxel2025.11 | 11.9 | 8.7 | 11.9 | 15 | — | — | — | — | — | |
| OccNeRFRep.=Voxel2025.11 | 10.4 | 6.9 | 10.3 | 14.1 | — | — | — | — | — | |
| SelfOccRep.=Voxel2025.11 | 9.6 | 6.7 | 9.6 | 12.5 | — | — | — | — | — | |
| BEVFormerBackbone=R101, Input Size=1600×9002025.06 | — | — | — | — | — | — | — | — | 39.1 | |
| FB-OccBackbone=R101, Input Size=1600×9002025.06 | — | — | — | — | — | — | — | — | 39.4 | |
| GraphGSOccBackbone=R101, Input Size=1600×9002025.06 | — | — | — | — | — | — | — | — | 42.6 | |
| OccNetBackbone=R101, Input Size=1600×9002025.06 | — | — | — | — | — | — | — | — | 36.8 | |
| RenderOccBackbone=R101, Input Size=1600×9002025.06 | — | — | — | — | — | — | — | — | 24.6 | |
| SparseOccBackbone=R101, Input Size=1600×9002025.06 | — | — | — | — | — | — | — | — | 30.6 |