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

This paper studies the problem of zero-shot sketch-based image retrieval(ZS-SBIR), which aims to use sketches from unseen categories as queries tomatch the images of the same category. Due to the large cross-modalitydiscrepancy, ZS-SBIR is still a challenging task and mimics realistic zero-shotscenarios. The key is to leverage transferable knowledge from the pre-trainedmodel to improve generalizability. Existing researchers often utilize thesimple fine-tuning training strategy or knowledge distillation from a teachermodel with fixed parameters, lacking efficient bidirectional knowledgealignment between student and teacher models simultaneously for bettergeneralization. In this paper, we propose a novel Symmetrical BidirectionalKnowledge Alignment for zero-shot sketch-based image retrieval (SBKA). Thesymmetrical bidirectional knowledge alignment learning framework is designed toeffectively learn mutual rich discriminative information between teacher andstudent models to achieve the goal of knowledge alignment. Instead of theformer one-to-one cross-modality matching in the testing stage, a one-to-manycluster cross-modality matching method is proposed to leverage the inherentrelationship of intra-class images to reduce the adverse effects of theexisting modality gap. Experiments on several representative ZS-SBIR datasets(Sketchy Ext dataset, TU-Berlin Ext dataset and QuickDraw Ext dataset) provethe proposed algorithm can achieve superior performance compared withstate-of-the-art methods.

Symmetrical Bidirectional Knowledge Alignment for Zero-Shot Sketch-Based Image Retrieval