Cross-view Referring Multi-Object Tracking

Cross-view Referring Multi-Object Tracking (CRMOT) is a complex task in the field of computer vision that aims to track multiple objects accurately by matching language descriptions and ensuring identity consistency of each object across different views. This task not only requires the system to understand natural language descriptions but also to have the capability of cross-view object recognition and tracking, to achieve efficient monitoring and intelligent analysis in multi-camera environments, which holds significant application value.