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

Despite deep convolutional neural networks' great success in objectclassification, it suffers from severe generalization performance drop underocclusion due to the inconsistency between training and testing data. Becauseof the large variance of occluders, our goal is a model trained onocclusion-free data while generalizable to occlusion conditions. In this work,we integrate prototypes, partial matching and top-down attention regulationinto deep neural networks to realize robust object classification underocclusion. We first introduce prototype learning as its regularizationencourages compact data clusters, which enables better generalization abilityunder inconsistent conditions. Then, attention map at intermediate layer basedon feature dictionary and activation scale is estimated for partial matching,which sifts irrelevant information out when comparing features with prototypes.Further, inspired by neuroscience research that reveals the important role offeedback connection for object recognition under occlusion, a top-down feedbackattention regulation is introduced into convolution layers, purposefullyreducing the contamination by occlusion during feature extraction stage. Ourexperiment results on partially occluded MNIST and vehicles from the PASCAL3D+dataset demonstrate that the proposed network significantly improves therobustness of current deep neural networks under occlusion. Our code will bereleased.

TDAPNet: Prototype Network with Recurrent Top-Down Attention for Robust Object Classification under Partial Occlusion