| Place | Computer Sciences Large Conference Room, Second Floor, University Services Center |
| Time | Mon & Wed 9:00 -10:20 |
| Instructor | Ross Beveridge |
This course surveys most recent topics in computer vision research. This semester will emphasize empirical evaluation, while at the same time surveying computer vision tasks that are both significant and timely. We will read a series of recent papers that illustrate different empirical approaches to the understanding of complex computer vision algorithms. These studies relate to tasks such as human face recognition, optimal geometric matching, texture classification, shape from shading, edge and ridge detection, vanishing point analysis and building extraction from aerial imagery. We will also read a series of recent survey articles as a means to develop a broader feel for some of the more mature sub-areas of computer vision. These areas will include statistical pattern recognition, handwriting recognition, medical image analysis and human identification.
The original course schedule has us meeting three times a week for 50 minutes. Seminars work better when there is a longer period of uninterrupted time, so it is my hope that this can be changed to twice a week for 75 minutes. Of course, this will only be done if alliterative time acceptable to all those interested in the course can be found. There is a tentative course schedule.
Class meetings will typically focus on one or several papers currently being read. Each student will read each assigned paper and write up a short synopsis. This synopsis must be submitted to me through email prior to the start of the corresponding class meeting. Typically, a student will be assigned to present the paper at least one week in advance of the class meeting. That student is responsible for providing a 15 to 25 minute overview of the paper and then leading a discussion. The presentation should be well prepared, including between 3 and 7 slides (powerpoint, slitex, or some equivalent). All other students in the course are expected to be active participants in the discussion.
The reading list is still being finalized. The course schedule includes the reading list, and this page includes information on where to obtain the paper, when it will be discussed, and who is presenting it.
Research is about developing, framing and answering questions. Part of the role of a seminar such as this one is to help develop this skill. Therefore, every student will do a semester project. Students are free to develop there own project concept. However, a series of suggested projects are being provided and students are encouraged to look these over carefully and give them due consideration.
Each project will have a phase I and phase II. Phase I is to be completed by mid-semester, and is itself expected to be a sizable amount of work (it is worth 25% of the total course grade). The phase I report will explain the project. In particular, it will contain 1) a clearly stated question (hypothesis) to be tested, 2) a clearly stated mechanism proposed for answering this question, and 3) a well written motivation explaining why the question is of interest and how it pertains to other work. It is also expected that prior to completing Phase I, either new code will have been developed to support phase II, and/or existing code will have been acquired, learned, and integrated into an experimental framework such that Phase II may be carried out.
Phase II essentially covers the work to be accomplished during the second half of the semester. It will include the conducting of the experiments designed and explained in Phase I. It can be expected to involve multiple iterations of experiment execution and refinement, since it is rare that all the relevant factors are perfectly accounted for in advance. At the end of the semester, each project will be thoroughly explained and presented in a written report. Finally, at the end of the semester, we will have a meeting where all the projects are presented orally.
Projects may be done individually or in team of up to three students. It should be understood that team will be expected to accomplish correspondingly more over the course of the semester.
| Synopses of reading assignments | 15 % |
| In class presentations | 15 % |
| In class participation | 10 % |
| Semester Project Phase I Report | 25 % |
| Semester Project Phase II Report | 30 % |
| Semester Project Phase II Presentation | 5 % |
| Total | 100 % |
There will be no late period for synopses, presentations, project proposals or project presentations.
You must have passed CS510 (or an equivalent graduate-level computer graphics course) or the experimental Introduction to Computer Vision class in order to take this course.
[TOP]
Last Updated, 9/11/2000, copyright, J. Ross Beveridge.