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

Learning-based image compression was shown to achieve a competitiveperformance with state-of-the-art transform-based codecs. This motivated thedevelopment of new learning-based visual compression standards such as JPEG-AI.Of particular interest to these emerging standards is the development oflearning-based image compression systems targeting both humans and machines.This paper is concerned with learning-based compression schemes whosecompressed-domain representations can be utilized to perform visual processingand computer vision tasks directly in the compressed domain. In our work, weadopt a learning-based compressed-domain classification framework forperforming visual recognition using the compressed-domain latent representationat varying bit-rates. We propose a novel feature adaptation module integratinga lightweight attention model to adaptively emphasize and enhance the keyfeatures within the extracted channel-wise information. Also, we design anadaptation training strategy to utilize the pretrained pixel-domain weights.For comparison, in addition to the performance results that are obtained usingour proposed latent-based compressed-domain method, we also present performanceresults using compressed but fully decoded images in the pixel domain as wellas original uncompressed images. The obtained performance results show that ourproposed compressed-domain classification model can distinctly outperform theexisting compressed-domain classification models, and that it can also yieldsimilar accuracy results with a much higher computational efficiency ascompared to the pixel-domain models that are trained using fully decodedimages.

DNN-Compressed Domain Visual Recognition with Feature Adaptation