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

Domain adaptive semantic segmentation enables robust pixel-wise understandingin real-world driving scenes. Source-free domain adaptation, as a morepractical technique, addresses the concerns of data privacy and storagelimitations in typical unsupervised domain adaptation methods. It utilizes awell-trained source model and unlabeled target data to achieve adaptation inthe target domain. However, in the absence of source data and target labels,current solutions cannot sufficiently reduce the impact of domain shift andfully leverage the information from the target data. In this paper, we proposean end-to-end source-free domain adaptation semantic segmentation method viaImportance-Aware and Prototype-Contrast (IAPC) learning. The proposed IAPCframework effectively extracts domain-invariant knowledge from the well-trainedsource model and learns domain-specific knowledge from the unlabeled targetdomain. Specifically, considering the problem of domain shift in the predictionof the target domain by the source model, we put forward an importance-awaremechanism for the biased target prediction probability distribution to extractdomain-invariant knowledge from the source model. We further introduce aprototype-contrast strategy, which includes a prototype-symmetric cross-entropyloss and a prototype-enhanced cross-entropy loss, to learn target intra-domainknowledge without relying on labels. A comprehensive variety of experiments ontwo domain adaptive semantic segmentation benchmarks demonstrates that theproposed end-to-end IAPC solution outperforms existing state-of-the-artmethods. Code will be made publicly available athttps://github.com/yihong-97/Source-free_IAPC.

Towards Source-free Domain Adaptive Semantic Segmentation via Importance-aware and Prototype-contrast Learning