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

Despite the large volume of face recognition datasets, there is a significantportion of subjects, of which the samples are insufficient and thusunder-represented. Ignoring such significant portion results in insufficienttraining data. Training with under-represented data leads to biased classifiersin conventionally-trained deep networks. In this paper, we propose acenter-based feature transfer framework to augment the feature space ofunder-represented subjects from the regular subjects that have sufficientlydiverse samples. A Gaussian prior of the variance is assumed across allsubjects and the variance from regular ones are transferred to theunder-represented ones. This encourages the under-represented distribution tobe closer to the regular distribution. Further, an alternating training regimenis proposed to simultaneously achieve less biased classifiers and a morediscriminative feature representation. We conduct ablative study to mimic theunder-represented datasets by varying the portion of under-represented classeson the MS-Celeb-1M dataset. Advantageous results on LFW, IJB-A and MS-Celeb-1Mdemonstrate the effectiveness of our feature transfer and training strategy,compared to both general baselines and state-of-the-art methods. Moreover, ourfeature transfer successfully presents smooth visual interpolation, whichconducts disentanglement to preserve identity of a class while augmenting itsfeature space with non-identity variations such as pose and lighting.

Feature Transfer Learning for Deep Face Recognition with Under-Represented Data