Table 6.
Point cloud object detection results [93,110]. mAPScanNet, mAPSUN RGB-D, and mAP3D results on ScanNet, SUN RGB-D, and KITTI datasets with only the ‘Car’ category.
| Model | Feature Extraction | mAPScanNet | mAPSUN RGB-D | mAP3D | ||
|---|---|---|---|---|---|---|
| Easy | Moderate | Hard | ||||
| FVNet [110] | PointNet | - | - | 65.43 | 57.34 | 51.85 |
| VoxelNet [108] | - | - | - | 81.97 | 65.46 | 62.85 |
| PointRCNN [107] | PointNet++, multi-scale grouping | - | - | 88.88 | 78.63 | 77.38 |
| F-PointNet [111] | PointNet++ | - | - | 81.20 | 70.39 | 62.19 |
| MVX-Net [109] | VoxelNet | - | - | 83.20 | 72.70 | 65.20 |
| Deep Hough voting model [112] | PointNet++ | 46.80 | 57.70 | - | - | - |