Due : Nov. 2, 2009

Computer Vision: The Computer Vision area exam takes the form of an independent research project agreed upon by the student and the faculty committee administering the exam. The goal is for the student to independently investigate a specific topic in a Computer Vision area. Relevant areas include Stereo, Human Motion Tracking, Image-based Immersive Environments and Visual Navigation for Robotics. Other Vision topics must be approved by the administering committee. The exam is composed of a survey of the relevant literature and a report on experiments performed to demonstrate or compare the most relevant approaches. Work should be primarily in the field of Computer Vision. A topic in a different area merely using images or image processing tools is not sufficient. Students attempting this exam must have completed CSCI 5722 Computer Vision.

Steps to completing this area exam include:

• Choosing a topic and writing a 1-2 page proposal and timeline which must be agreed to (signed) by one of the committee members.
• Writing a literature survey which briefly describes relevant work in the topic area.
• Hands on experiments with the leading techniques or an original method, including demonstration and discussion of the performance of these approaches on images or sequences agreed on in the original proposal. Comparisons should be meaningful and quantitative. Students are encouraged to find implementations available on the web to facilitate these experiments.

The area exam process is expected to begin well before the final due date and extend over several months, with the restriction that the final paper is due on the area exam date (Nov. 2). Students must notify the committee of their intention to participate in the area exam at least 30 days prior to the due date (preferably earlier), and must complete a proposal which is acceptable to the committee.

Students who do not submit their papers by the due date will not have taken the exam. Students who submit papers deemed inadequate by the committee will fail the exam.

Normally the faculty committee administering this area exam will include Jane Mulligan, Jim Martin and Michael Mozer.