Skip to main content

Human Tracking and Re-Identification

Hwang_human-tracking-and-re-identificationWhen a single camera is used to monitor and count a scene full of people, it’s common for a person who’s already been tracked by the system to re-enter the camera’s field of view multiple times. This throws off the counting and results in incorrect statistics of distinct people. Based on our tracking trajectories and the holistic and regional appearance features, this project is successfully able to continuously re-identify a person who has already been tracked to avoid repeated counting.