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

Transformer-based Self-supervised Representation Learning methods learngeneric features from unlabeled datasets for providing useful networkinitialization parameters for downstream tasks. Recently, self-supervisedlearning based upon masking local surface patches for 3D point cloud data hasbeen under-explored. In this paper, we propose masked Autoencoders in 3D pointcloud representation learning (abbreviated as MAE3D), a novel autoencodingparadigm for self-supervised learning. We first split the input point cloudinto patches and mask a portion of them, then use our Patch Embedding Module toextract the features of unmasked patches. Secondly, we employ patch-wise MAE3DTransformers to learn both local features of point cloud patches and high-levelcontextual relationships between patches and complete the latentrepresentations of masked patches. We use our Point Cloud Reconstruction Modulewith multi-task loss to complete the incomplete point cloud as a result. Weconduct self-supervised pre-training on ShapeNet55 with the point cloudcompletion pre-text task and fine-tune the pre-trained model on ModelNet40 andScanObjectNN (PB\_T50\_RS, the hardest variant). Comprehensive experimentsdemonstrate that the local features extracted by our MAE3D from point cloudpatches are beneficial for downstream classification tasks, soundlyoutperforming state-of-the-art methods ($93.4\%$ and $86.2\%$ classificationaccuracy, respectively).

Masked Autoencoders in 3D Point Cloud Representation Learning