Advances in 3D Point Cloud Learning

The field of 3D point cloud learning is rapidly evolving, with a focus on developing innovative methods to improve the performance and efficiency of deep learning models. Recent research has explored the use of geometry-aware approaches, such as latent geometric prototypes and point transformers, to enhance feature generation and alignment in 3D point cloud segmentation and classification tasks. Additionally, there is a growing interest in developing methods that can generalize well across different point cloud domains, including those with varying levels of occlusion and missing points. Another area of research is the application of 3D point cloud learning in medical domains, including disease diagnosis and treatment, where hierarchical feature learning frameworks have shown promising results. Noteworthy papers in this area include:

  • CMIP-CIL, which proposes a cross-modal benchmark for image-point class incremental learning and achieves state-of-the-art results.
  • 3D-PointZshotS, which introduces a geometry-aware zero-shot segmentation framework that enhances both feature generation and alignment using latent geometric prototypes.

Sources

Boosting the Class-Incremental Learning in 3D Point Clouds via Zero-Collection-Cost Basic Shape Pre-Training

CMIP-CIL: A Cross-Modal Benchmark for Image-Point Class Incremental Learning

Adaptive Decision Boundary for Few-Shot Class-Incremental Learning

Leveraging Point Transformers for Detecting Anatomical Landmarks in Digital Dentistry

3D-PointZshotS: Geometry-Aware 3D Point Cloud Zero-Shot Semantic Segmentation Narrowing the Visual-Semantic Gap

DG-MVP: 3D Domain Generalization via Multiple Views of Point Clouds for Classification

Computer-Aided Design of Personalized Occlusal Positioning Splints Using Multimodal 3D Data

Hierarchical Feature Learning for Medical Point Clouds via State Space Model

Built with on top of