3D Semantic Occupancy Prediction on SurroundOcc (val)
28.9mIoUGaussianOcc3D
Evaluation results
| Method | Links | |||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| GaussianOcc3DModality=C&L, Query Form=3D Gaussian, Query Number=256002026.01 | 28.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GaussianOcc3DModality=Camera & LiDAR2026.01 | 28.9 | — | — | 28.7 | 1,680 | 34.8 | 38.5 | 19.8 | 2,310 | 25.7 | 1,390 | 22.7 | 30.9 | 43.3 | 25.3 | 29.1 | 30.1 | 34.7 | 45.1 | |
| Co-OccModality=C&L, Query Form=3D Voxel, Query Number=100×100×82026.01 | 27.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GaussianFormer3DModality=C&L, Query Form=3D Gaussian, Query Number=256002026.01 | 27.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Co-OccModality=Camera & LiDAR2026.01 | 27.1 | — | — | 28.1 | 1,610 | 34 | 37.2 | 17 | 2,160 | 20.8 | 1,590 | 21.9 | 28.7 | 42.3 | 25.4 | 29.1 | 28.6 | 28.2 | 38 | |
| GaussianFormer3DModality=Camera & LiDAR2026.01 | 27.1 | — | — | 26.9 | 1,580 | 32.7 | 36.1 | 18.6 | 2,170 | 24.1 | 1,300 | 21.3 | 29 | 40.6 | 23.7 | 27.3 | 28.2 | 32.6 | 42.3 | |
| OccFusionModality=Camera & LiDAR2026.01 | 26.8 | — | — | 26.6 | 1,830 | 32.9 | 35.8 | 19.3 | 2,210 | 24.4 | 1,770 | 21.4 | 29.6 | 39 | 21.9 | 24.9 | 26.7 | 28.5 | 40 | |
| GaussianOcc3DModality=C&L, Query Form=3D Gaussian, Query Number=128002026.01 | 26.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| M-CONetModality=C&L, Query Form=3D Voxel, Query Number=100×100×82026.01 | 24.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| M-CONetModality=Camera & LiDAR2026.01 | 24.7 | — | — | 24.8 | 1,300 | 31.6 | 34.8 | 14.6 | 1,800 | 20 | 1,470 | 20 | 26.6 | 39.2 | 22.8 | 26.1 | 26 | 26 | 37.1 | |
| GaussianFormer3DModality=C&L, Query Form=3D Gaussian, Query Number=128002026.01 | 24.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| StreamOccLatency (ms)=84, Memory (MB)=2,7882025.03 | 23.4 | 21 | 33.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GaussianWorldLatency (ms)=228, Memory (MB)=7,0302025.03 | 21.9 | 19 | 33 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| QuadricFormerLatency (ms)=179, Memory (MB)=2,5632025.03 | 21.1 | 17.3 | 32.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GaussianFormer-2Modality=Camera2026.01 | 20.8 | — | — | 21.4 | 1,340 | 28.5 | 30.8 | 10.9 | 1,580 | 13.6 | 1,050 | 14 | 22.9 | 40.6 | 24.4 | 26.1 | 24.3 | 13.8 | 22 | |
| GaussianFormer-2Latency (ms)=451, Memory (MB)=4,5352025.03 | 20.8 | 18.3 | 31.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GaussianFormer-2Modality=C, Query Form=3D Gaussian, Query Number=256002026.01 | 20.3 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SurroundOccModality=Camera2026.01 | 20.3 | — | — | 20.6 | 1,170 | 28.1 | 30.9 | 10.7 | 1,510 | 14.1 | 1,210 | 14.4 | 22.3 | 37.3 | 23.7 | 24.5 | 22.8 | 14.9 | 21.9 | |
| SurroundOccLatency (ms)=344, Memory (MB)=5,4912025.03 | 20.3 | 16 | 31.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GaussianFormer-2Modality=C, Query Form=3D Gaussian, Query Number=64002026.01 | 19.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GaussianFormer-2Modality=C, Query Form=3D Gaussian, Query Number=128002026.01 | 19.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FB-OccModality=Camera2026.01 | 19.6 | — | — | 20.6 | 1,130 | 26.9 | 29.8 | 10.4 | 1,360 | 13.7 | 1,140 | 11.5 | 20.6 | 38.2 | 21.5 | 24.6 | 22.7 | 14.8 | 21.6 | |
| GaussianFormerModality=Camera2026.01 | 19.1 | — | — | 19.5 | 1,130 | 26.1 | 29.8 | 10.5 | 1,380 | 12.6 | 870 | 12.7 | 21.6 | 39.6 | 23.3 | 24.5 | 23 | 9.6 | 19.1 | |
| GaussianFormerLatency (ms)=372, Memory (MB)=6,2292025.03 | 19.1 | 16.7 | 29.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| OccFormerModality=Camera2026.01 | 19 | — | — | 18.7 | 1,040 | 23.9 | 30.3 | 10.3 | 1,420 | 13.6 | 1,010 | 12.5 | 20.8 | 38.8 | 19.8 | 24.2 | 22.2 | 13.5 | 21.4 | |
| C-CONetModality=Camera2026.01 | 18.4 | — | — | 18.6 | 1,000 | 26.4 | 27.4 | 8.6 | 1,570 | 13.3 | 970 | 10.9 | 20.2 | 33 | 20.7 | 21.4 | 21.8 | 14.7 | 21.3 | |
| L-CONetModality=LiDAR2026.01 | 17.7 | — | — | 19.2 | 400 | 15.1 | 26.9 | 6.2 | 380 | 6.8 | 600 | 14.1 | 13.1 | 39.7 | 19.1 | 24 | 23.9 | 25.1 | 35.7 | |
| TPVFormerModality=Camera2026.01 | 17.1 | — | — | 16 | 530 | 23.9 | 27.3 | 9.8 | 870 | 7.1 | 520 | 11 | 19.2 | 38.9 | 21.3 | 24.3 | 23.2 | 11.7 | 20.8 | |
| TPVFormerLatency (ms)=320, Memory (MB)=5,1002025.03 | 17.1 | 13.4 | 30.9 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| BEVFormerModality=Camera2026.01 | 16.8 | — | — | 14.2 | 660 | 23.5 | 28.3 | 8.7 | 1,080 | 6.6 | 410 | 11.2 | 17.8 | 37.3 | 18 | 22.9 | 22.2 | 13.8 | 22.2 | |
| LMSCNetModality=LiDAR2026.01 | 14.9 | — | — | 13.1 | 450 | 14.7 | 22.1 | 12.6 | 420 | 7.2 | 710 | 12.2 | 11.5 | 26.3 | 14.3 | 21.1 | 15.2 | 18.5 | 34.2 | |
| MonoSceneModality=Camera2026.01 | 7.3 | — | — | 4 | 40 | 8 | 8 | 2.9 | 30 | 1.2 | 70 | 4 | 4.4 | 27.7 | 5.2 | 15.1 | 11.3 | 9 | 14.9 | |
| ALOcc-GFBackbone=ResNet-50, Input Size=900 x 1600, FPS=0.9, Memory=11857, Temporal Scheme=GDFusion2025.04 | 0.255 | 0.225 | 0.382 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ALOcc*Backbone=ResNet-50, Input Size=900 x 1600, FPS=1, Memory=11117, temporal_fusion=false2025.04 | 0.24 | 0.217 | 0.347 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ALOcc-mini-GFBackbone=ResNet-50, Input Size=900 x 1600, FPS=5.4, Memory=3441, Temporal Scheme=GDFusion2025.04 | 0.231 | 0.203 | 0.346 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GaussianWorldBackbone=ResNet-101, Input Size=900 x 1600, FPS=4.4, Memory=70302025.04 | 0.221 | 0.197 | 0.334 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ALOcc-mini*Backbone=ResNet-50, Input Size=900 x 1600, FPS=5.8, Memory=2869, temporal_fusion=false2025.04 | 0.215 | 0.195 | 0.315 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| BEVDetOcc-GFBackbone=ResNet-50, Input Size=900 x 1600, FPS=1.1, Memory=8447, Temporal Scheme=GDFusion2025.04 | 0.208 | 0.162 | 0.353 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| SurroundOccBackbone=ResNet-101, Input Size=900 x 16002025.04 | 0.203 | 0.184 | 0.315 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GaussianFormerBackbone=ResNet-101, Input Size=900 x 1600, FPS=2.7, Memory=62292025.04 | 0.191 | 0.173 | 0.298 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| BEVDetOccBackbone=ResNet-50, Input Size=900 x 1600, FPS=1.1, Memory=81912025.04 | 0.175 | 0.141 | 0.292 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| TPVFormerBackbone=ResNet-101, Input Size=900 x 1600, FPS=2.9, Memory=290002025.04 | 0.171 | 0.14 | 0.309 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| BEVFormerBackbone=ResNet-101, Input Size=900 x 1600, FPS=3.3, Memory=251002025.04 | 0.168 | 0.142 | 0.305 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — |