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

The task of point cloud upsampling aims to acquire dense and uniform pointsets from sparse and irregular point sets. Although significant progress hasbeen made with deep learning models, state-of-the-art methods requireground-truth dense point sets as the supervision, which makes them limited tobe trained under synthetic paired training data and not suitable to be underreal-scanned sparse data. However, it is expensive and tedious to obtain largenumbers of paired sparse-dense point sets as supervision from real-scannedsparse data. To address this problem, we propose a self-supervised point cloudupsampling network, named SPU-Net, to capture the inherent upsampling patternsof points lying on the underlying object surface. Specifically, we propose acoarse-to-fine reconstruction framework, which contains two main components:point feature extraction and point feature expansion, respectively. In thepoint feature extraction, we integrate the self-attention module with the graphconvolution network (GCN) to capture context information inside and among localregions simultaneously. In the point feature expansion, we introduce ahierarchically learnable folding strategy to generate upsampled point sets withlearnable 2D grids. Moreover, to further optimize the noisy points in thegenerated point sets, we propose a novel self-projection optimizationassociated with uniform and reconstruction terms as a joint loss to facilitatethe self-supervised point cloud upsampling. We conduct various experiments onboth synthetic and real-scanned datasets, and the results demonstrate that weachieve comparable performances to state-of-the-art supervised methods.

SPU-Net: Self-Supervised Point Cloud Upsampling by Coarse-to-Fine Reconstruction with Self-Projection Optimization