MRCN: A Novel Modality Restitution and Compensation Network for Visible-Infrared Person Re-identification

Visible-infrared person re-identification (VI-ReID), which aims to search identities across different spectra, is a challenging task due large cross-modality discrepancy between visible and infrared images. The key reduce the filter out identity-irrelevant interference effectively learn modality-invariant representations. In this paper, we propose novel Modality Restitution Compensation Network (MRCN) narrow gap two modalities. Specifically, first modality by using Instance Normalization (IN) layers. Next, influence of IN layers on removing discriminative information differences, Module (MRM) (MCM) respectively distill modality-irrelevant modality-relevant features from removed information. Then, are used restitute normalized features, while compensate for other modality. Furthermore, better disentangle Center-Quadruplet Causal (CQC) loss encourage network features. Extensive experiments conducted validate superiority our method SYSU-MM01 RegDB datasets. More remarkably, achieves 95.1% in terms Rank-1 89.2% mAP dataset.