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

Periocular biometric, the peripheral area of the ocular, is a collaborativealternative to the face, especially when the face is occluded or masked.However, in practice, sole periocular biometric capture the least salientfacial features, thereby lacking discriminative information, particularly inwild environments. To address these problems, we transfer discriminatoryinformation from the face to support the training of a periocular network byusing knowledge distillation. Specifically, we leverage face images forperiocular embedding learning, but periocular alone is utilized for identityidentification or verification. To enhance periocular embeddings by faceeffectively, we proposeConsistent Knowledge Distillation (CKD) that imposesconsistency between face and periocular networks across prediction and featurelayers. We find that imposing consistency at the prediction layer enables (1)extraction of global discriminative relationship information from face imagesand (2) effective transfer of the information from the face network to theperiocular network. Particularly, consistency regularizes the prediction unitsto extract and store profound inter-class relationship information of faceimages. (3) The feature layer consistency, on the other hand, makes theperiocular features robust against identity-irrelevant attributes. Overall, CKDempowers the sole periocular network to produce robust discriminativeembeddings for periocular recognition in the wild. We theoretically andempirically validate the core principles of the distillation mechanism in CKD,discovering that CKD is equivalent to label smoothing with a novelsparsity-oriented regularizer that helps the network prediction to capture theglobal discriminative relationship. Extensive experiments reveal that CKDachieves state-of-the-art results on standard periocular recognition benchmarkdatasets.

Periocular Embedding Learning with Consistent Knowledge Distillation from Face