Self-Supervised Representation Learning for CAD

Dataset Analysis

Proposes self-supervised representation learning for CAD B-Rep geometry using a hybrid implicit/explicit surface, improving few-shot learning and achieving SOTA on benchmarks.