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/

Zendy Plus

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

Zendy Tools

Zendy Open

Person re-identification (Re-ID) aims to match the image frames which containthe same person in the surveillance videos. Most of the Re-ID algorithmsconduct supervised training in some small labeled datasets, so directlydeploying these trained models to the real-world large camera networks may leadto a poor performance due to underfitting. The significant difference betweenthe source training dataset and the target testing dataset makes it challengingto incrementally optimize the model. To address this challenge, we propose anovel solution by transforming the unlabeled images in the target domain to fitthe original classifier by using our proposed similarity preserved generativeadversarial networks model, SimPGAN. Specifically, SimPGAN adopts thegenerative adversarial networks with the cycle consistency constraint totransform the unlabeled images in the target domain to the style of the sourcedomain. Meanwhile, SimPGAN uses the similarity consistency loss, which ismeasured by a siamese deep convolutional neural network, to preserve thesimilarity of the transformed images of the same person. Comprehensiveexperiments based on multiple real surveillance datasets are conducted, and theresults show that our algorithm is better than the state-of-the-artcross-dataset unsupervised person Re-ID algorithms.

Cross-dataset Person Re-Identification Using Similarity Preserved Generative Adversarial Networks