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

Existing person re-identification (re-ID) research mainly focuses onpedestrian identity matching across cameras in adjacent areas. However, inreality, it is inevitable to face the problem of pedestrian identity matchingacross long-distance scenes. The cross-camera pedestrian samples collected fromlong-distance scenes often have no positive samples. It is extremelychallenging to use cross-camera negative samples to achieve cross-regionpedestrian identity matching. Therefore, a novel domain-adaptive person re-IDmethod that focuses on cross-camera consistent discriminative feature learningunder the supervision of unpaired samples is proposed. This method mainlyincludes category synergy co-promotion module (CSCM) and cross-cameraconsistent feature learning module (CCFLM). In CSCM, a task-specific featurerecombination (FRT) mechanism is proposed. This mechanism first groups featuresaccording to their contributions to specific tasks. Then an interactivepromotion learning (IPL) scheme between feature groups is developed andembedded in this mechanism to enhance feature discriminability. Since thecontrol parameters of the specific task model are reduced after division bytask, the generalization ability of the model is improved. In CCFLM,instance-level feature distribution alignment and cross-camera identityconsistent learning methods are constructed. Therefore, the supervised modeltraining is achieved under the style supervision of the target domain byexchanging styles between source-domain samples and target-domain samples, andthe challenges caused by the lack of cross-camera paired samples are solved byutilizing cross-camera similar samples. In experiments, three challengingdatasets are used as target domains, and the effectiveness of the proposedmethod is demonstrated through four experimental settings.

Domain-adaptive Person Re-identification without Cross-camera Paired Samples