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 Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

To deal with the exhausting annotations, self-supervised representationlearning from unlabeled point clouds has drawn much attention, especiallycentered on augmentation-based contrastive methods. However, specificaugmentations hardly produce sufficient transferability to high-level tasks ondifferent datasets. Besides, augmentations on point clouds may also changeunderlying semantics. To address the issues, we propose a simple but efficientaugmentation fusion contrastive learning framework to combine dataaugmentations in Euclidean space and feature augmentations in feature space. Inparticular, we propose a data augmentation method based on sampling and graphgeneration. Meanwhile, we design a data augmentation network to enable acorrespondence of representations by maximizing consistency between augmentedgraph pairs. We further design a feature augmentation network that encouragesthe model to learn representations invariant to the perturbations using anencoder perturbation. We comprehensively conduct extensive objectclassification experiments and object part segmentation experiments to validatethe transferability of the proposed framework. Experimental results demonstratethat the proposed framework is effective to learn the point cloudrepresentation in a self-supervised manner, and yields state-of-the-art resultsin the community. The source code is publicly available at:https://zhiyongsu.github.io/Project/AFSRL.html.

Joint Data and Feature Augmentation for Self-Supervised Representation Learning on Point Clouds