Learning to track objects by synthesizing object tracks.




large-scale object tracking databases are hard to annotate and thus are lacking. This project will study data synthesis methods using graphics engines or generative adversarial networks to generate object tracking datasets, such that a robust object tracker can be trained. My group has rich experience in developing object recognition, tracking and data synthesis systems. A perspective student is expected to have research experience in computer vision, graphics or machine learning.


Updated:  10 August 2021/Responsible Officer:  Dean, CECS/Page Contact:  CECS Marketing