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

Visible-infrared person re-identification (V-I ReID) seeks to match images ofindividuals captured over a distributed network of RGB and IR cameras. The taskis challenging due to the significant differences between V and I modalities,especially under real-world conditions, where images are corrupted by, e.g,blur, noise, and weather. Indeed, state-of-art V-I ReID models cannot leveragecorrupted modality information to sustain a high level of accuracy. In thispaper, we propose an efficient model for multimodal V-I ReID -- namedMultimodal Middle Stream Fusion (MMSF) -- that preserves modality-specificknowledge for improved robustness to corrupted multimodal images. In addition,three state-of-art attention-based multimodal fusion models are adapted toaddress corrupted multimodal data in V-I ReID, allowing to dynamically balanceeach modality importance. Recently, evaluation protocols have been proposed toassess the robustness of ReID models under challenging real-world scenarios.However, these protocols are limited to unimodal V settings. For realisticevaluation of multimodal (and cross-modal) V-I person ReID models, we proposenew challenging corrupted datasets for scenarios where V and I cameras areco-located (CL) and not co-located (NCL). Finally, the benefits of our Maskingand Local Multimodal Data Augmentation (ML-MDA) strategy are explored toimprove the robustness of ReID models to multimodal corruption. Our experimentson clean and corrupted versions of the SYSU-MM01, RegDB, and ThermalWORLDdatasets indicate the multimodal V-I ReID models that are more likely toperform well in real-world operational conditions. In particular, our ML-MDA isan important strategy for a V-I person ReID system to sustain high accuracy androbustness when processing corrupted multimodal images. Also, our multimodalReID model MMSF outperforms every method under CL and NCL camera scenarios.

Fusion for Visual-Infrared Person ReID in Real-World Surveillance Using Corrupted Multimodal Data