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

With the success of large-scale visual-language pretraining models and thewide application of image-text retrieval in industry areas, reducing the modelsize and streamlining their terminal-device deployment have become urgentlynecessary. The mainstream model structures for image-text retrieval aresingle-stream and dual-stream, both aiming to close the semantic gap betweenvisual and textual modalities. Dual-stream models excel at offline indexing andfast inference, while single-stream models achieve more accurate cross-modelalignment by employing adequate feature fusion. We propose a multi-teachercross-modality alignment distillation (MCAD) technique to integrate theadvantages of single-stream and dual-stream models. By incorporating the fusedsingle-stream features into the image and text features of the dual-streammodel, we formulate new modified teacher features and logits. Then, we conductboth logit and feature distillation to boost the capability of the studentdual-stream model, achieving high retrieval performance without increasinginference complexity. Extensive experiments demonstrate the remarkableperformance and high efficiency of MCAD on image-text retrieval tasks.Furthermore, we implement a mobile CLIP model on Snapdragon clips with only 93Mrunning memory and 30ms search latency, without apparent performancedegradation of the original large CLIP.

MCAD: Multi-teacher Cross-modal Alignment Distillation for efficient image-text retrieval