Grant: Canada Research Chairs

The proposed research is in computer vision, a field that studies the use of computers to automatically perceive and understand visual content in order to make predictions or help make decisions. A key component of vision is learning, allowing acquisition of sophisticated models of the visual world from large image/video corpora. This program focuses on developing, implementing and evaluating novel approaches for semantic visual understanding at scale. The program will address fundamental challenges, including learning from limited and multi-modal data (e.g., images + descriptive text), by developing novel neural architectures endowed with memory, semantics and structured learning objectives.