GeoMask3D: Geometrically Informed Mask Selection for Self-Supervised Point Cloud Learning in 3D
This work addresses the challenge of enhancing feature representation in 3D point cloud learning for applications in computer vision, though it appears incremental as it builds upon existing MAE frameworks.
The paper tackles the problem of inefficient self-supervised learning for point clouds by introducing GeoMask3D, a geometrically informed mask selection strategy that replaces random masking in Masked Auto Encoders, resulting in improved performance on downstream tasks like classification and few-shot learning.
We introduce a pioneering approach to self-supervised learning for point clouds, employing a geometrically informed mask selection strategy called GeoMask3D (GM3D) to boost the efficiency of Masked Auto Encoders (MAE). Unlike the conventional method of random masking, our technique utilizes a teacher-student model to focus on intricate areas within the data, guiding the model's focus toward regions with higher geometric complexity. This strategy is grounded in the hypothesis that concentrating on harder patches yields a more robust feature representation, as evidenced by the improved performance on downstream tasks. Our method also presents a complete-to-partial feature-level knowledge distillation technique designed to guide the prediction of geometric complexity utilizing a comprehensive context from feature-level information. Extensive experiments confirm our method's superiority over State-Of-The-Art (SOTA) baselines, demonstrating marked improvements in classification, and few-shot tasks.