Advanced Learning for Automatic Object Detection and Tracking in UAV Imagery

Recently there has been a growing adoption of Unmanned Aerial Vehicles (UAVs) in many applications. Robust object detection and tracking algorithms are required in many UAV scenarios. However, existing object detection and tracking approaches are not specifically designed for UAV and therefore do not have satisfactory performance in UAV. In this MITACS cluster project, by working with the industrial partner, we will develop a set of improved object detection and tracking algorithms for UAV. In particular, we will tackle the following sub-topics: 1.) weakly supervised object localization; 2.) visual object tracking and anomaly detection; 3.) partial face detection; and 4.) low-resolution object detection. The algorithms will be integrated into the partnerÂ’s products, thereby significantly enhancing their competitiveness in the market, and contributing more to Canadian economy and job market.

Intern: 
Jiannan Zheng
Pegah Kharazmi
Faculty Supervisor: 
Rabab Ward
Province: 
British Columbia
Partner University: 
Program: