Random Walks in Self-supervised Learning for Triangular Meshes

This study addresses the challenge of self-supervised learning for 3D mesh analysis. It presents an new approach that uses random walks as a form of data augmentation to generate diverse representations of mesh surfaces. Further- more, it employs a combination of contrastive and clustering losses. The contrastive learning framework maximizes simi-larity between augmented instances of the same mesh while minimizing similarity between different meshes. We inte-grate this with a clustering loss, enhancing class distinction across training epochs and mitigating training variance. Our model’s effectiveness is evaluated using mean Average Pre-cision (mAP) scores and a supervised SVM linear classifier on extracted features, demonstrating its potential for various downstream tasks such as object classification and shape retrieval.

M.Sc. student under the supervision of Prof. Ayellet Tal.