Section: Bilateral Contracts and Grants with Industry
Participants : Karteek Alahari, Cordelia Schmid.
We received a Google Faculty Research Award in 2015. The objective is to interpret video semantically in the presence of weak supervision. We will focus on answering questions such as who is in the scene, what they are doing, and when exactly did they perform their action(s). We propose to develop models for detection and recognition of objects and actions learned from minimally annotated training data.