Research Article
DANC-Net: Dual-Attention and Negative Constraint Network for Point Cloud Classification
Table 2
Classification performance on ModelNet10. (The top three accuracies are highlighted by bold, underline, and italic.)
| | Methods | Input | Points (k) | mA (%) | OA (%) |
| | VoxNet (IROS 2015) | Points | 1 | — | 92.0 | | 3DShapeNets (CVPR 2016) | Points | 1 | — | 83.5 | | PointNet (CVPR 2017) | Points | 1 | 94.2 | 94.4 | | PointNet++ (CVPR 2017) | Points + normal | 5 | 94.7 | 94.9 | | Kd-Net (ICCV 2017) | Points | 32 | 93.5 | 94.0 | | DGCNN (TOG 2019) | Points | 1 | 94.8 | 94.9 | | A-CNN (CVPR 2019) | Points + normal | 1 | 94.4 | 94.6 | | Ours | Points + normal | 1 | 95.4 | 95.5 |
|
|
Input and points represent the input data type and the number of sampling points, respectively.
|
