3D Object Detection on nuScenes (val) (mAP and NDS)
75mAPGaussianFusion-T
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GaussianFusion-Ttemporal_extension=true2026.07 | 75 | 77.6 | |
| SparseLIF-Ttemporal_extension=true2026.07 | 74.7 | 77.5 | |
| BEVFusion4Dtemporal_extension=true2026.07 | 72 | 73.5 | |
| FusionFormertemporal_extension=true2026.07 | 71.4 | 74.1 | |
| SToRe3D-1/2Backbone=ViT-L, Image Width (W)=800, Sparsity Level (S)=1/2, FPS=2.72026.05 | 53.3 | 61.8 | |
| ToC3D-FastBackbone=ViT-L, Image Width (W)=800, Sparsity Level (S)=1/2, FPS=2.52026.05 | 52.3 | 61 | |
| SToRe3D-1/3Backbone=ViT-L, Image Width (W)=800, Sparsity Level (S)=1/3, FPS=3.52026.05 | 52.3 | 60.9 | |
| StreamPETRBackbone=ViT-L, Image Width (W)=800, FPS=2.22026.05 | 52.1 | 60.8 | |
| SToRe3D-1/10Backbone=ViT-L, Image Width (W)=800, Sparsity Level (S)=1/10, FPS=5.22026.05 | 52.1 | 60.7 | |
| ToC3D-FasterBackbone=ViT-L, Image Width (W)=800, Sparsity Level (S)=1/3, FPS=3.12026.05 | 51.7 | 60.9 | |
| Sparse4Dv2Backbone=R101, Image Width (W)=14082026.05 | 50.5 | 59.8 | |
| SparseBEVBackbone=R101, Image Width (W)=14082026.05 | 50.1 | 59.2 | |
| StreamPETRBackbone=ViT-B, Image Width (W)=800, FPS=6.12026.05 | 49.7 | 58.4 | |
| SToRe3D-1/2Backbone=ViT-B, Image Width (W)=800, Sparsity Level (S)=1/2, FPS=8.22026.05 | 49.3 | 58.1 | |
| SToRe3D-1/3Backbone=ViT-B, Image Width (W)=800, Sparsity Level (S)=1/3, FPS=10.62026.05 | 48.9 | 57.8 | |
| StreamPETRBackbone=R101, Image Width (W)=1408, FPS=7.02026.05 | 48.6 | 57.8 | |
| SOLOFusionBackbone=R101, Image Width (W)=14082026.05 | 48.3 | 58.2 | |
| StreamPETRBackbone=V2-99, Image Width (W)=800, FPS=13.52026.05 | 48.2 | 57.1 | |
| SToRe3D-1/10Backbone=ViT-B, Image Width (W)=800, Sparsity Level (S)=1/10, FPS=17.72026.05 | 47.9 | 57.1 | |
| ToC3D-FastBackbone=ViT-B, Image Width (W)=800, Sparsity Level (S)=1/2, FPS=6.62026.05 | 46 | 56.2 | |
| ToC3D-FasterBackbone=ViT-B, Image Width (W)=800, Sparsity Level (S)=1/3, FPS=7.32026.05 | 45.3 | 55.9 | |
| Sparse4DBackbone=R101-DCN, Image Width (W)=1600, FPS=4.72026.05 | 44.4 | 55 | |
| BEVFormerBackbone=R101-DCN, Image Width (W)=1600, FPS=3.32026.05 | 41.6 | 51.7 |