3

Towards Efficient Point Cloud Classification via Critical Point Aware Pooling with Adaptive Learnable Threshold
We propose Critical Point Aware Pooling (CPAPool), which learns the importance of points with adaptive thresholds to down-sample point clouds, reducing computation while improving classification accuracy on ModelNet10 and ModelNet40.
Tong Sang, Qiyu Wei 魏启宇, Xiaofeng Zou, Zeng Zeng, Xulei Yang
Towards Efficient Point Cloud Classification via Critical Point Aware Pooling with Adaptive Learnable Threshold