English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Although recent point cloud analysis achieves impressive progress, theparadigm of representation learning from a single modality gradually meets itsbottleneck. In this work, we take a step towards more discriminative 3D pointcloud representation by fully taking advantages of images which inherentlycontain richer appearance information, e.g., texture, color, and shade.Specifically, this paper introduces a simple but effective point cloudcross-modality training (PointCMT) strategy, which utilizes view-images, i.e.,rendered or projected 2D images of the 3D object, to boost point cloudanalysis. In practice, to effectively acquire auxiliary knowledge from viewimages, we develop a teacher-student framework and formulate the cross modallearning as a knowledge distillation problem. PointCMT eliminates thedistribution discrepancy between different modalities through novel feature andclassifier enhancement criteria and avoids potential negative transfereffectively. Note that PointCMT effectively improves the point-onlyrepresentation without architecture modification. Sufficient experiments verifysignificant gains on various datasets using appealing backbones, i.e., equippedwith PointCMT, PointNet++ and PointMLP achieve state-of-the-art performance ontwo benchmarks, i.e., 94.4% and 86.7% accuracy on ModelNet40 and ScanObjectNN,respectively. Code will be made available athttps://github.com/ZhanHeshen/PointCMT.

Let Images Give You More:Point Cloud Cross-Modal Training for Shape Analysis