Research Article
Research on 3D Point Cloud Object Detection Algorithm for Autonomous Driving
Table 2
Comparison of algorithm performance at a moderate level.
| Class | Method | Modality | 3DmAP |
| Car | F-PointNet [16] | | 70.92% | AVOD-FPN [18] | | 74.44% | ContFuse [20] | RGB + LiDAR | 73.25% | MV3D [15] | | 62.68% | Ours | | 75.02% |
| Pedestrian | F-PointNet [16] | | 61.32% | AVOD-FPN [18] | | 58.8% | ContFuse [20] | RGB + LiDAR | — | MV3D [15] | | — | Ours | | 62.51% |
| Cyclist | F-PointNet [16] | | 56.49% | AVOD-FPN [18] | | 49.7% | ContFuse [20] | RGB + LiDAR | — | MV3D [15] | | — | Ours | | 57.26% |
|
|