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

Processing 3D data efficiently has always been a challenge. Spatialoperations on large-scale point clouds, stored as sparse data, require extracost. Attracted by the success of transformers, researchers are usingmulti-head attention for vision tasks. However, attention calculations intransformers come with quadratic complexity in the number of inputs and missspatial intuition on sets like point clouds. We redesign set transformers inthis work and incorporate them into a hierarchical framework for shapeclassification and part and scene segmentation. We propose our local attentionunit, which captures features in a spatial neighborhood. We also computeefficient and dynamic global cross attentions by leveraging sampling andgrouping at each iteration. Finally, to mitigate the non-heterogeneity of pointclouds, we propose an efficient Multi-Scale Tokenization (MST), which extractsscale-invariant tokens for attention operations. The proposed hierarchicalmodel achieves state-of-the-art shape classification in mean accuracy andyields results on par with the previous segmentation methods while requiringsignificantly fewer computations. Our proposed architecture predictssegmentation labels with around half the latency and parameter count of theprevious most efficient method with comparable performance. The code isavailable at https://github.com/YigeWang-WHU/CloudAttention.

CloudAttention: Efficient Multi-Scale Attention Scheme For 3D Point Cloud Learning