Research Article
DANC-Net: Dual-Attention and Negative Constraint Network for Point Cloud Classification
Table 4
Overall accuracy in % on ScanObjectNN. (The top three accuracies are highlighted by bold, underline, and italic.)
| Methods | OBJ_BG | PB_T25 | PB_T25_R | PB_T50_R | PB_T50_RS | Mean of OA |
| PointNet | 79.0 | 74.5 | 73.2 | 69.3 | 67.8 | 72.7 | PointNet++ | 83.5 | 85.4 | 82.8 | 80.9 | 78.7 | 82.2 | DGCNN | 85.3 | 85.7 | 83.8 | 80.9 | 81.0 | 83.3 | A-CNN | 85.1 | 83.2 | 83.5 | 81.8 | 81.4 | 83.0 | Ours | 86.4 | 84.3 | 84.0 | 82.7 | 81.8 | 83.8 |
|
|