Counterfactual Intervention Feature Transfer for Visible-Infrared Person Re-identification
Researchers have developed a new method called Counterfactual Intervention Feature Transfer (CIFT) to improve visible-infrared person re-identification. This technique addresses issues like modality imbalance between training and testing phases, and suboptimal graph topology learning. CIFT utilizes a Homogeneous and Heterogeneous Feature Transfer module to mitigate the modality gap and a Counterfactual Relation Intervention component to enhance the reliability of the graph topology structure. Experiments show CIFT surpasses existing state-of-the-art methods on standard benchmarks. AI
IMPACT Enhances person re-identification accuracy by addressing modality gaps and improving graph topology learning.