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

Unsupervised Domain Adaptation (UDA) aims to enhance the generalization ofthe learned model to other domains. The domain-invariant knowledge istransferred from the model trained on labeled source domain, e.g., video game,to unlabeled target domains, e.g., real-world scenarios, saving annotationexpenses. Existing UDA methods for semantic segmentation usually focus onminimizing the inter-domain discrepancy of various levels, e.g., pixels,features, and predictions, for extracting domain-invariant knowledge. However,the primary intra-domain knowledge, such as context correlation inside animage, remains underexplored. In an attempt to fill this gap, we propose aunified pixel- and patch-wise self-supervised learning framework, called PiPa,for domain adaptive semantic segmentation that facilitates intra-imagepixel-wise correlations and patch-wise semantic consistency against differentcontexts. The proposed framework exploits the inherent structures ofintra-domain images, which: (1) explicitly encourages learning thediscriminative pixel-wise features with intra-class compactness and inter-classseparability, and (2) motivates the robust feature learning of the identicalpatch against different contexts or fluctuations. Extensive experiments verifythe effectiveness of the proposed method, which obtains competitive accuracy onthe two widely-used UDA benchmarks, i.e., 75.6 mIoU on GTA to Cityscapes and68.2 mIoU on Synthia to Cityscapes. Moreover, our method is compatible withother UDA approaches to further improve the performance without introducingextra parameters.

PiPa: Pixel- and Patch-wise Self-supervised Learning for Domain Adaptative Semantic Segmentation