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

Despite its importance, unsupervised domain adaptation (UDA) on LiDARsemantic segmentation is a task that has not received much attention from theresearch community. Only recently, a completion-based 3D method has beenproposed to tackle the problem and formally set up the adaptive scenarios.However, the proposed pipeline is complex, voxel-based and requires multi-stageinference, which inhibits it for real-time inference. We propose a rangeimage-based, effective and efficient method for solving UDA on LiDARsegmentation. The method exploits class prototypes from the source domain topseudo label target domain pixels, which is a research direction showing goodperformance in UDA for natural image semantic segmentation. Applying suchapproaches to LiDAR scans has not been considered because of the severe domainshift and lack of pre-trained feature extractor that is unavailable in theLiDAR segmentation setup. However, we show that proper strategies, includingreconstruction-based pre-training, enhanced prototypes, and selective pseudolabeling based on distance to prototypes, is sufficient enough to enable theuse of prototypical approaches. We evaluate the performance of our method onthe recently proposed LiDAR segmentation UDA scenarios. Our method achievesremarkable performance among contemporary methods.

Enhanced Prototypical Learning for Unsupervised Domain Adaptation in LiDAR Semantic Segmentation