eduzhai > Applied Sciences > Engineering >

Info3D Representation Learning on 3D Objects using Mutual Information Maximization and Contrastive Learning

  • Save

... pages left unread,continue reading

Document pages: 17 pages

Abstract: A major endeavor of computer vision is to represent, understand and extractstructure from 3D data. Towards this goal, unsupervised learning is a powerfuland necessary tool. Most current unsupervised methods for 3D shape analysis usedatasets that are aligned, require objects to be reconstructed and suffer fromdeteriorated performance on downstream tasks. To solve these issues, we proposeto extend the InfoMax and contrastive learning principles on 3D shapes. We showthat we can maximize the mutual information between 3D objects and their "chunks " to improve the representations in aligned datasets. Furthermore, wecan achieve rotation invariance in SO${(3)}$ group by maximizing the mutualinformation between the 3D objects and their geometric transformed versions.Finally, we conduct several experiments such as clustering, transfer learning,shape retrieval, and achieve state of art results.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×