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

The generalization capability of unsupervised domain adaptation can mitigatethe need for extensive pixel-level annotations to train semantic segmentationnetworks by training models on synthetic data as a source withcomputer-generated annotations. Entropy-based adversarial networks are proposedto improve source domain prediction; however, they disregard significantexternal information, such as edges, which have the potential to identify anddistinguish various objects within an image accurately. To address this issue,we introduce a novel approach to domain adaptation, leveraging the synergy ofinternal and external information within entropy-based adversarial networks. Inthis approach, we enrich the discriminator network with edge-predictedprobability values within this innovative framework to enhance the clarity ofclass boundaries. Furthermore, we devised a probability-sharing network thatintegrates diverse information for more effective segmentation. Incorporatingobject edges addresses a pivotal aspect of unsupervised domain adaptation thathas frequently been neglected in the past -- the precise delineation of objectboundaries. Conventional unsupervised domain adaptation methods usually centeraround aligning feature distributions and may not explicitly model objectboundaries. Our approach effectively bridges this gap by offering clearguidance on object boundaries, thereby elevating the quality of domainadaptation. Our approach undergoes rigorous evaluation on the establishedunsupervised domain adaptation benchmarks, specifically in adapting SYNTHIA$\rightarrow$ Cityscapes and SYNTHIA $\rightarrow$ Mapillary. Experimentalresults show that the proposed model attains better performance thanstate-of-the-art methods. The superior performance across differentunsupervised domain adaptation scenarios highlights the versatility androbustness of the proposed method.

Robust Unsupervised Domain Adaptation by Retaining Confident Entropy via Edge Concatenation