3D scene understanding on Replica (Target View)
73LSeg mIoUFeature-3DGS
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Feature-3DGSFeed Forward=false, Input Pose=true, #G=199K, Memories=141.8MB2025.12 | 73 | 93.6 | 35.7 | 0.972 | 0.044 | 42.1 | 68.6 | 35.9 | 0.972 | 0.045 | |
| CF³Feed Forward=false, Input Pose=true, #G=10K, Memories=7.1MB2025.12 | 66.3 | 91.8 | 27.49 | 0.906 | 0.132 | 38 | 65.4 | 27.49 | 0.906 | 0.132 | |
| C3GFeed Forward=true, Input Pose=false, #G=2K, Memories=4.1MB2025.12 | 63 | 89.3 | 25.43 | 0.818 | 0.173 | 41.6 | 69.2 | 25 | 0.809 | 0.182 | |
| LSeg / MaskCLIP*Feed Forward=false2025.12 | 61.8 | 88.7 | — | — | — | 41.2 | 66.8 | — | — | — | |
| LSMFeed Forward=true, Input Pose=false, #G=131K, Memories=61.5MB2025.12 | 60 | 82.3 | 21.86 | 0.753 | 0.213 | 24.1 | 41.1 | 17.01 | 0.637 | 0.377 |