IMPORTANT NOTE: The class times and location listed on CMPT's course central are incorrect. The correct times (from go.sfu.ca) are as above, 2h on Mondays, no Friday class. The location of the class is TASC1 8002.
In this course we will study approaches to tackle the problem of human activity recognition. The grand goal in this area of computer vision is to build systems that can automatically find human figures in either images or video sequences and determine what action they are performing. The recognition of human figures and actions is an important problem, with many practical applications. This technology is directly applicable to human-computer interaction, image and video retrieval and search, security and surveillance, video motion capture and automated vehicle driver assistance systems.
This seminar-style course is intended for students who wish to learn about the state of the art in vision-based human activity recognition. We will cover the major approaches in the literature, for the most part by reading and discussing research papers. Students will be expected to have, or obtain, the background knowledge in computer vision and machine learning that will be needed to understand these papers. CMPT 726 (Machine Learning) and a computer vision / image processing course would provide useful, though not essential, background knowledge. If you are concerned about your readiness for this course, please ask.