Recognition Problems in Computer Vision

OBJECTIVE/DESCRIPTION:

Human beings are adept at a variety of "recognition" problems, such as
recognizing everyday objects or the actions and identities of familiar
people.  In this course we will explore computer vision techniques
that have the grand goal of giving computers these same abilities.
Classical techniques and recent original research addressing these
problems will be presented.  Students in graphics and machine learning
should be interested by the applications to image-based rendering and
the substantial problem domains respectively.

TOPICS:

- Shape matching
- Face detection and recognition
- Handwritten digit recognition
- Generic 3d object recognition
- Activity recognition
- Scene/location recognition
- Detection of human figures
- Texture recognition
- Motion synthesis

GRADING:

To be discussed in the first week of classes.  The grade will mostly
be determined by a course project, with accompanying report and
presentation.  There will also be reading assignments, and a couple of
small homework assignments.

TEXTBOOKS:

None

PREREQUISITES/COREQUISITES:

Interest in computer vision.  An undergraduate course in computer
vision or image processing would be an asset, but not required.
Please feel free to contact me if you have any concerns about your
suitability for taking this course.