Winter 2012
Instructor: Lihi Zelnik-Manor
Time: Monday 14:30-16:30
Location: Meyer 351
Office hours: Thu. 14:00-15:00
Welcome to the seminar on “Advanced Topics in Computer Vision”. In this course, every week we will study a different topic, presented by one or two students. Each participant in the course is required to present one or two topics as a seminar. Additionally, every week all students are expected to read one of the relevant papers and submit a one-paragraph summary, prior to the seminar.
The topics we will cover include some of latest work in this area of research.
About the class
Grading policy
Since this is a seminar, attending the class is obligatory. If you cannot attend a certain lecture, please send me an email in advance.
The final grade will be based on:
1) The quality of your presentation
2) Sending it to me in time (Sunday morning, the week before your talk)
3) Attendance
4) Sending me a one-paragraph summary every week
First assignment:
By Saturday November 3rd, send me a list of your preferred topics (with priorities). The topics will be allocated to students on the first-come-first-serve strategy (the site will be continuously updated).
You may also suggest a topic of interest to you.
You should wait for my formal confirmation before starting to work on a topic.
Your talk
The presentation of a topic should include:
- What are the main computer vision challenges in this field?
- What are the main research problems people are working on?
- What do you think people should be working on?
- A summary of one or two relevant papers
- Your personal evaluation of the papers.
When evaluating a paper you should discuss the strengths and weaknesses of the presented approaches by considering the following aspects:
- Does the paper make reasonable assumptions?
- How novel is the solution?
- Is the solution technically sound?
- How convincing is the experimental evaluation?
- Writing level: is the paper clearly written? Is it self-contained?
- Most importantly: any ideas for interesting future directions?