Point Cloud Classification
Point cloud classification aims to automatically assign labels to 3D point cloud data, enabling applications in robotics, autonomous driving, and more. Current research emphasizes improving efficiency and robustness through techniques like knowledge distillation, self-supervised learning, and novel architectures such as MLPs, Transformers, and state space models, often incorporating geometric features or addressing challenges like noise, occlusion, and domain generalization. These advancements are crucial for deploying reliable and efficient point cloud processing in resource-constrained environments and real-world scenarios where data quality is variable.
Papers
Efficient Point Cloud Classification via Offline Distillation Framework and Negative-Weight Self-Distillation Technique
Qiang Zheng, Chao Zhang, Jian Sun
PMT-MAE: Dual-Branch Self-Supervised Learning with Distillation for Efficient Point Cloud Classification
Qiang Zheng, Chao Zhang, Jian Sun
SA-MLP: Enhancing Point Cloud Classification with Efficient Addition and Shift Operations in MLP Architectures
Qiang Zheng, Chao Zhang, Jian Sun