Louisiana State University Department of Electrical and Computer Engineering EE 7700 - Advanced Topics in Computer Vision Spring 2007 Class Schedule: MWF 11:40pm - 12:30pm CEBA - 3140 Instructor: Dr. Bahadir K. Gunturk Room: ECE - 225 Tel: (225) 578 - 5621 Office Hours: MWF 8:30am-11:30pm Syllabus: In this course, we will study selected topics in computer vision. Prerequisite: Background on computer vision and programming in Matlab Format: Each week, a paper is assigned. Students are required to study the paper before coming to the class. The paper will be discussed during the class. Students are sometimes asked to implement a paper/idea and write a report. Outline of a typical technical paper: - Abstract - Introduction (to the problem) - Related work - Method (solution approach) - Experiments - Conclusions Tips on paper reading: First read once: - Understand the problem and the approach - Do not try to understand the details - Skip the literature survey (do not go and read every paper mentioned) Read more: - Understand the details - Think about the strengths and weaknesses, and how to improve the ideas - Think about how to implement the paper How to evaluate/review a paper: • What is the main contribution of the paper?
• Is the abstract appropriate and adequate digest of the paper?
• Does the introduction clearly state the background?
• Is the paper's coverage of the chosen topic comprehensive?
• Is the performance of the proposed system/algorithm rigorously characterized?
• Does the paper provide enough information to replicate their experiments?
• Is software made available to the community?
• Is the list of references adequate?
• Are the relative lengths of sections appropriate for the material covered?
• Are there too-many/too-few figures? Are all figures appropriately captioned?
• How readable is the paper (regardless of technical content)? How to write a paper: Follow the outline as shown above. Determine the main points of the paper and explain/demonstrate them clearly. Online resources: Google, IEEE Xplore (connect to LSU library website --> Indexes/Databases --> Engineering --> IEEE Xplore) The tentative grading policy will be as follows: Projects: 40% Midterm: 20% Final: 30% Discussion and Attendance: 10%