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 Tools annual

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 monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Point cloud, an efficient 3D object representation, has become popular withthe development of depth sensing and 3D laser scanning techniques. It hasattracted attention in various applications such as 3D tele-presence,navigation for unmanned vehicles and heritage reconstruction. The understandingof point clouds, such as point cloud segmentation, is crucial in exploiting theinformative value of point clouds for such applications. Due to theirregularity of the data format, previous deep learning works often convertpoint clouds to regular 3D voxel grids or collections of images before feedingthem into neural networks, which leads to voluminous data and quantizationartifacts. In this paper, we instead propose a regularized graph convolutionalneural network (RGCNN) that directly consumes point clouds. Leveraging onspectral graph theory, we treat features of points in a point cloud as signalson graph, and define the convolution over graph by Chebyshev polynomialapproximation. In particular, we update the graph Laplacian matrix thatdescribes the connectivity of features in each layer according to thecorresponding learned features, which adaptively captures the structure ofdynamic graphs. Further, we deploy a graph-signal smoothness prior in the lossfunction, thus regularizing the learning process. Experimental results on theShapeNet part dataset show that the proposed approach significantly reduces thecomputational complexity while achieving competitive performance with the stateof the art. Also, experiments show RGCNN is much more robust to both noise andpoint cloud density in comparison with other methods. We further apply RGCNN topoint cloud classification and achieve competitive results on ModelNet40dataset.

RGCNN: Regularized Graph CNN for Point Cloud Segmentation