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

Existing 3D instance segmentation methods are predominated by the bottom-updesign -- manually fine-tuned algorithm to group points into clusters followedby a refinement network. However, by relying on the quality of the clusters,these methods generate susceptible results when (1) nearby objects with thesame semantic class are packed together, or (2) large objects with looselyconnected regions. To address these limitations, we introduce ISBNet, a novelcluster-free method that represents instances as kernels and decodes instancemasks via dynamic convolution. To efficiently generate high-recall anddiscriminative kernels, we propose a simple strategy named Instance-awareFarthest Point Sampling to sample candidates and leverage the local aggregationlayer inspired by PointNet++ to encode candidate features. Moreover, we showthat predicting and leveraging the 3D axis-aligned bounding boxes in thedynamic convolution further boosts performance. Our method set newstate-of-the-art results on ScanNetV2 (55.9), S3DIS (60.8), and STPLS3D (49.2)in terms of AP and retains fast inference time (237ms per scene on ScanNetV2).The source code and trained models are available athttps://github.com/VinAIResearch/ISBNet.

ISBNet: a 3D Point Cloud Instance Segmentation Network with Instance-aware Sampling and Box-aware Dynamic Convolution