GeoMask3D: Geometrically Informed Mask Selection for Self-Supervised Point Cloud Learning in 3D

TL;DR

GeoMask3D introduces a geometrically informed mask selection strategy for self-supervised point cloud learning, improving feature robustness and performance on downstream tasks by focusing on complex regions during training.

Contribution

The paper proposes GeoMask3D, a novel mask selection method using geometric complexity, and a feature-level knowledge distillation technique for better self-supervised point cloud learning.

Findings

Outperforms SOTA baselines in classification tasks

Enhances few-shot learning performance

Improves feature robustness through geometric focus

Abstract

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.