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

This paper challenges the cross-domain semantic segmentation task, aiming toimprove the segmentation accuracy on the unlabeled target domain withoutincurring additional annotation. Using the pseudo-label-based unsuperviseddomain adaptation (UDA) pipeline, we propose a novel and effective MultipleFusion Adaptation (MFA) method. MFA basically considers three parallelinformation fusion strategies, i.e., the cross-model fusion, temporal fusionand a novel online-offline pseudo label fusion. Specifically, theonline-offline pseudo label fusion encourages the adaptive training to payadditional attention to difficult regions that are easily ignored by offlinepseudo labels, therefore retaining more informative details. While the othertwo fusion strategies may look standard, MFA pays significant efforts to raisethe efficiency and effectiveness for integration, and succeeds in injecting allthe three strategies into a unified framework. Experiments on two widely usedbenchmarks, i.e., GTA5-to-Cityscapes and SYNTHIA-to-Cityscapes, show that ourmethod significantly improves the semantic segmentation adaptation, and sets upnew state of the art (58.2% and 62.5% mIoU, respectively). The code will beavailable at https://github.com/KaiiZhang/MFA.

Multiple Fusion Adaptation: A Strong Framework for Unsupervised Semantic Segmentation Adaptation