3D Object Detection on CA-1M
40.9AP@0.25 IoUCuTR (RGB-D)
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| CuTR (RGB-D)Input Paradigm=2D image-based, Input Modality=RGB-D, Post-processing=None, Detection limit=100, RGB Resolution=1024x768, Depth Resolution=256x1922024.12 | 40.9 | 62.3 | 12.7 | 29.1 | |
| FCAFInput Paradigm=3D point-based, Post-processing=NMS, Detection limit=1002024.12 | 29.3 | 49.5 | 11.2 | 22.6 | |
| TR3D + FFInput Paradigm=3D point-based, Post-processing=NMS, Detection limit=1002024.12 | 24.8 | 52.9 | 4.7 | 21 | |
| TR3DInput Paradigm=3D point-based, Post-processing=NMS, Detection limit=1002024.12 | 22 | 51.9 | 4.4 | 20 | |
| CuTR (RGB only)Input Paradigm=2D image-based, Input Modality=RGB only, Post-processing=None, Detection limit=100, RGB Resolution=1024x7682024.12 | 13.5 | 35.4 | 2.4 | 12.9 | |
| ImVoxelNet (RGB only)Input Paradigm=3D point-based, Input Modality=RGB only, Post-processing=NMS, Detection limit=1002024.12 | 10.1 | 22.8 | 2.3 | 6.3 | |
| Cube R-CNN (RGB only)Input Paradigm=2D image-based, Input Modality=RGB only, Post-processing=NMS, Detection limit=1002024.12 | 4.6 | 20.1 | 1 | 4.7 |