The main objective of our project is to develop an algorithm to track one or more objects across a video sequence, and give an alert if one of the objects changes trajectory sharply. Ideally, we would like to learn the normal behavior of the object and then be able to detect unexpected behavior. In addition, we would like to infer the future trajectory of objects so that we can build an agent that avoids collisions. Similarly, we’re interested in using machine learning techniques to classify different movement patterns. For instance, can an agent detect if something is chasing it?
Robert Giaquinto and Thomas Ringstrom