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 Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Text-based person retrieval aims to identify specific individuals within animage database using textual descriptions. Due to the high cost of annotationand privacy protection, researchers resort to synthesized data for the paradigmof pretraining and fine-tuning. However, these generated data often exhibitdomain biases in both images and textual annotations, which largely compromisethe scalability of the pre-trained model. Therefore, we introduce adomain-agnostic pretraining framework based on Cross-modality AdaptiveMeta-Learning (CAMeL) to enhance the model generalization capability duringpretraining to facilitate the subsequent downstream tasks. In particular, wedevelop a series of tasks that reflect the diversity and complexity ofreal-world scenarios, and introduce a dynamic error sample memory unit tomemorize the history for errors encountered within multiple tasks. To furtherensure multi-task adaptation, we also adopt an adaptive dual-speed updatestrategy, balancing fast adaptation to new tasks and slow weight updates forhistorical tasks. Albeit simple, our proposed model not only surpasses existingstate-of-the-art methods on real-world benchmarks, including CUHK-PEDES,ICFG-PEDES, and RSTPReid, but also showcases robustness and scalability inhandling biased synthetic images and noisy text annotations. Our code isavailable at https://github.com/Jahawn-Wen/CAMeL-reID.

CAMeL: Cross-modality Adaptive Meta-Learning for Text-based Person Retrieval