Research Article
Pillar-Based 3D Object Detection from Point Cloud with Multiattention Mechanism
Table 3
Results on the KITTI test average orientation similarity (AOS) detection benchmark.
| Model | Car | Pedestrian | Cyclist | Easy (%) | Mod. (%) | Hard (%) | mAP (%) | Easy (%) | Mod. (%) | Hard (%) | mAP (%) | Easy (%) | Mod. (%) | Hard (%) | mAP (%) |
| PointPillars | 90.19 | 88.76 | 86.38 | 88.44 | 32.39 | 31.41 | 29.84 | 31.21 | 82.43 | 68.16 | 61.96 | 70.85 | PointPillars + point-wise attention | 90.53 | 88.75 | 87.25 | 88.84 | 50.18 | 46.40 | 43.62 | 46.73 | 84.49 | 69.75 | 65.53 | 73.25 | PointPillars + SOPA | 90.56 | 88.77 | 87.26 | 88.86 | 54.67 | 50.85 | 47.43 | 50.97 | 85.09 | 73.38 | 69.71 | 76.06 |
|
|