Point Discriminative Learning for Data-efficient 3D Point Cloud Analysis
This addresses the challenge of reducing the need for large labeled datasets in 3D point cloud analysis, which is important for applications in robotics and autonomous systems, but it appears incremental as it builds on existing self-supervision techniques.
The paper tackles the problem of data-efficient 3D point cloud analysis by proposing PointDisc, a method that uses self-supervision with a point discrimination loss to improve classification and segmentation, achieving benefits in experiments on tasks like 3D object classification and semantic segmentation.
3D point cloud analysis has drawn a lot of research attention due to its wide applications. However, collecting massive labelled 3D point cloud data is both time-consuming and labor-intensive. This calls for data-efficient learning methods. In this work we propose PointDisc, a point discriminative learning method to leverage self-supervisions for data-efficient 3D point cloud classification and segmentation. PointDisc imposes a novel point discrimination loss on the middle and global level features produced by the backbone network. This point discrimination loss enforces learned features to be consistent with points belonging to the corresponding local shape region and inconsistent with randomly sampled noisy points. We conduct extensive experiments on 3D object classification, 3D semantic and part segmentation, showing the benefits of PointDisc for data-efficient learning. Detailed analysis demonstrate that PointDisc learns unsupervised features that well capture local and global geometry.