SL3D: Self-supervised-Self-labeled 3D Recognition
This addresses the scalability issue in 3D recognition for researchers and practitioners by reducing reliance on expensive manual annotations, though it appears incremental as it builds on existing self-supervised and clustering techniques.
The paper tackles the problem of manually annotating 3D data for recognition tasks by proposing SL3D, a self-supervised and self-labeled framework that generates pseudo-labeled data through clustering and feature learning, achieving effectiveness in classification, detection, and segmentation.
Deep learning has attained remarkable success in many 3D visual recognition tasks, including shape classification, object detection, and semantic segmentation. However, many of these results rely on manually collecting densely annotated real-world 3D data, which is highly time-consuming and expensive to obtain, limiting the scalability of 3D recognition tasks. Thus, we study unsupervised 3D recognition and propose a Self-supervised-Self-Labeled 3D Recognition (SL3D) framework. SL3D simultaneously solves two coupled objectives, i.e., clustering and learning feature representation to generate pseudo-labeled data for unsupervised 3D recognition. SL3D is a generic framework and can be applied to solve different 3D recognition tasks, including classification, object detection, and semantic segmentation. Extensive experiments demonstrate its effectiveness. Code is available at https://github.com/fcendra/sl3d.