| 方法 | 年份 | 类别 | Car (IoU = 0.7) | Pedestrian (IoU = 0.5) | Cyclist (IoU = 0.5) | mAP |
| PointPillars [19] | 2019 | 基于2D图像衍化而来的 3D点云数据增强方法 | 74.11 | 45.70 | 62.59 | 59.20 |
| PointRCNN [30] | 2019 | 针对3D点云设计的增强方法 | 76.67 | 43.28 | 62.37 | 60.77 |
| PV-RCNN [34] | 2020 | 针对3D点云设计的增强方法 | 82.83 | 45.25 | 66.65 | 64.91 |
| PPBA [46] | 2020 | 混合与创新型数据增强方法 | 77.67 | 46.09 | 65.58 | 63.11 |
| PA-AUG [43] | 2021 | 混合与创新型数据增强方法 | 83.08 | 61.59 | 75.59 | 73.42 |
| PointDrop [31] | 2021 | 针对3D点云设计的增强方法 | 77.99 | 61.31 | 67.10 | 68.80 |
| Pattern-Aware [32] | 2021 | 针对3D点云设计的增强方法 | 86.49 | 59.41 | 76.79 | 74.23 |
| CA-aug [40] | 2023 | 针对3D点云设计的增强方法 | 86.61 | 60.39 | 78.84 | 75.28 |