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

Current point cloud segmentation architectures suffer from limited long-rangefeature modeling, as they mostly rely on aggregating information with localneighborhoods. Furthermore, in order to learn point features at multiplescales, most methods utilize a data-agnostic sampling approach to decrease thenumber of points after each stage. Such sampling methods, however, oftendiscard points for small objects in the early stages, leading to inadequatefeature learning. We believe these issues are can be mitigated by introducingexplicit geometry clues as guidance. To this end, we propose GeoSpark, aPlug-in module that incorporates Geometry clues into the network to Spark upfeature learning and downsampling. GeoSpark can be easily integrated intovarious backbones. For feature aggregation, it improves feature modeling byallowing the network to learn from both local points and neighboring geometrypartitions, resulting in an enlarged data-tailored receptive field.Additionally, GeoSpark utilizes geometry partition information to guide thedownsampling process, where points with unique features are preserved whileredundant points are fused, resulting in better preservation of key pointsthroughout the network. We observed consistent improvements after addingGeoSpark to various backbones including PointNet++, KPConv, andPointTransformer. Notably, when integrated with Point Transformer, our GeoSparkmodule achieves a 74.7% mIoU on the ScanNetv2 dataset (4.1% improvement) and71.5% mIoU on the S3DIS Area 5 dataset (1.1% improvement), ranking top on bothbenchmarks. Code and models will be made publicly available.

GeoSpark: Sparking up Point Cloud Segmentation with Geometry Clue