サインイン
インサイト - Point Cloud Classification
Leveraging Local Neighborhood Features to Enhance 3D Point Cloud Classification
Incorporating radius-normalized distance and directional vectors as additional local neighborhood features can significantly improve the classification accuracy of 3D point cloud models, particularly on real-world datasets.
Efficient Point Cloud Classification with Dual-Branch Self-Supervised Learning and Knowledge Distillation
The PMT-MAE framework, featuring a dual-branch architecture that integrates Transformer and MLP components, along with a two-stage distillation strategy, achieves high accuracy and efficiency in point cloud classification tasks.
