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

Optical flow has achieved great success under clean scenes, but suffers fromrestricted performance under foggy scenes. To bridge the clean-to-foggy domaingap, the existing methods typically adopt the domain adaptation to transfer themotion knowledge from clean to synthetic foggy domain. However, these methodsunexpectedly neglect the synthetic-to-real domain gap, and thus are erroneouswhen applied to real-world scenes. To handle the practical optical flow underreal foggy scenes, in this work, we propose a novel unsupervised cumulativedomain adaptation optical flow (UCDA-Flow) framework: depth-association motionadaptation and correlation-alignment motion adaptation. Specifically, wediscover that depth is a key ingredient to influence the optical flow: thedeeper depth, the inferior optical flow, which motivates us to design adepth-association motion adaptation module to bridge the clean-to-foggy domaingap. Moreover, we figure out that the cost volume correlation shares similardistribution of the synthetic and real foggy images, which enlightens us todevise a correlation-alignment motion adaptation module to distill motionknowledge of the synthetic foggy domain to the real foggy domain. Note thatsynthetic fog is designed as the intermediate domain. Under this unifiedframework, the proposed cumulative adaptation progressively transfers knowledgefrom clean scenes to real foggy scenes. Extensive experiments have beenperformed to verify the superiority of the proposed method.

Unsupervised Cumulative Domain Adaptation for Foggy Scene Optical Flow