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

The challenge in fine-grained visual categorization lies in how to explorethe subtle differences between different subclasses and achieve accuratediscrimination. Previous research has relied on large-scale annotated data andpre-trained deep models to achieve the objective. However, when only a limitedamount of samples is available, similar methods may become less effective.Diffusion models have been widely adopted in data augmentation due to theiroutstanding diversity in data generation. However, the high level of detailrequired for fine-grained images makes it challenging for existing methods tobe directly employed. To address this issue, we propose a novel approach termedthe detail reinforcement diffusion model~(DRDM), which leverages the richknowledge of large models for fine-grained data augmentation and comprises twokey components including discriminative semantic recombination (DSR) andspatial knowledge reference~(SKR). Specifically, DSR is designed to extractimplicit similarity relationships from the labels and reconstruct the semanticmapping between labels and instances, which enables better discrimination ofsubtle differences between different subclasses. Furthermore, we introduce theSKR module, which incorporates the distributions of different datasets asreferences in the feature space. This allows the SKR to aggregate thehigh-dimensional distribution of subclass features in few-shot FGVC tasks, thusexpanding the decision boundary. Through these two critical components, weeffectively utilize the knowledge from large models to address the issue ofdata scarcity, resulting in improved performance for fine-grained visualrecognition tasks. Extensive experiments demonstrate the consistent performancegain offered by our DRDM.

Detail Reinforcement Diffusion Model: Augmentation Fine-Grained Visual Categorization in Few-Shot Conditions