Identity Consistency Construction for Visible-Infrared Person Re-identification in Cloud Environment

The real-scene deployment of the dual-mode surveillance camera system enables the collection of different modality images such as visible images and infrared images, ensuring continuous monitoring during both day and night. However, the obvious disparity between different modality images poses significant challenges for person re-identification tasks. Furthermore, the applications of cross-modal person re-ID methods are based on large amounts of data. Due to limited storage resources, it is challenging to ensure models work properly on local devices. To tackle this problem, we propose a novel cloud-based identity consistency learning method in this paper. This method employs a dual-classifier collaborative strategy to leverage the extracted features from different perspectives. By integrating these features, the model can extract more complementary information from both visible and infrared images. This integrated design contributes to the enhancement of the model's performance. It can be seen that the method's superiority has been demonstrated through the outstanding results on the SYSU-MM01 dataset.