3D Semantic Scene Completion on SemanticKITTI 25.6m range (val)
57.69IoU (Overall)VoxFormer-T
Evaluation Results
| Method | Links | |||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| VoxFormer-TTemporal Information=Current + Previous 4 images, Input Modality=Camera2023.02 | 57.69 | 69.95 | 76.7 | 18.42 | 37.46 | 2.87 | 1.24 | 10.38 | 10.61 | 3.5 | 3.92 | 0 | 66.15 | 23.96 | 34.53 | 0.76 | 29.45 | 11.15 | 38.07 | 12.75 | 39.61 | 15.56 | 8.09 | |
| VoxFormer-STemporal Information=Current image only, Input Modality=Camera2023.02 | 57.54 | 70.85 | 75.39 | 16.48 | 35.24 | 1.48 | 1.1 | 7.47 | 4.98 | 3.31 | 7.14 | 0 | 65.74 | 18.49 | 33.2 | 1.54 | 24.09 | 10.63 | 34.68 | 10.64 | 35.08 | 10.63 | 6.29 | |
| LMSCNet*Input Modality=RGB-inferred (pseudo LiDAR)2023.02 | 54.89 | 82.21 | 62.29 | 14.13 | 35.41 | 0 | 0 | 3.49 | 0 | 0 | 0 | 0 | 67.56 | 13.22 | 34.2 | 0 | 27.83 | 4.42 | 33.32 | 3.01 | 41.51 | 4.94 | 0 | |
| SSCNet*Input Modality=RGB-inferred (pseudo LiDAR)2023.02 | 53.2 | 59.13 | 84.15 | 14.55 | 31.09 | 0 | 0 | 8.49 | 4.55 | 0 | 0.02 | 0 | 65.78 | 13.35 | 32.84 | 0.01 | 24.59 | 4.53 | 33.17 | 8.53 | 38.46 | 8.33 | 2.65 | |
| MonoSceneInput Modality=Camera2023.02 | 38.55 | 51.96 | 59.91 | 12.22 | 24.64 | 0.23 | 0.2 | 13.84 | 2.13 | 1.37 | 1 | 0 | 57.11 | 18.6 | 27.58 | 2 | 15.97 | 7.37 | 19.68 | 2.57 | 31.59 | 3.79 | 2.54 |