Computer Science 153
Computer Vision
Assignments/Grading Policies, Fall 2008

Assignments

This course consists of six projects. There are no midterm/final exams. The first five are approximately two-week projects, and the final one provides a chance to explore an algorithm/application - and implement one - of your own choosing.

Project 6    Single-view modeling    (Due by Wednesday, November 26)
Proposal    for the final project    (Due Monday, November 17)
Project 5    Eigenfaces    (Due Sunday, November 9)
Project 4    Image mosaicking, part 2: autostitching    (Due Friday, Oct 24)
Project 3    Image mosaicking, part 1: 2d visual geometry    (Due Friday, Oct 10)
Project 2    Seam finder!    (Due Fri, Sep 26)
Project 1    Spamfinder!    (Due Fri, Sep 12)

The last assignment is a final project. It consists of an investigation, analysis, write-up, and class presentation of a vision-based system or technology of your own design. There will be a project proposal due on Nov. 13 (above). In addition, there will be a presentation of about 30 minutes in class during one of the semester's final two weeks. We will conclude on the day of CS 153's exam with each individual or team demoing their project to the group.

Assignment Grading and Submission

The projects will be due about every other week on Fridays.
For each project, you are required to (1) create a webpage or a wiki page summarizing your results -- I would encourage you to take advantage of this as motivation to create a coursework "showoff page" to which you might point potential employers/grad schools/others.

In addition, you are required to (2) demo your software on the due date to me. I'm available pretty much any time on Fridays (before 4:30pm!) for this.

Teamwork You are encouraged to work in teams of up to three people on these projects; I'd consider two people an optimal size team. You may join or dissolve teams on a project-by-project basis. The CS honesty code applies to teamwork: the effort must be equally shared by all of the team members. Discussions are absolutely encouraged between teams, but the work of the project must be each group's own.

Your webpage "submission" will not need to have a link to the source code -- rather, I will simply copy an archive of your demoed project to a central course location at the time of the demo.

Late days are certainly possible, but they come at a price: namely, the presentation of a 10-15 minute technical overview of a vision algorithm or application of the day at the start of one of the lectures. If you'll need to use a late day or so on a project, let me know so that we can set up a time for your demo (and the "vision of the day")!

Course Grading

For each of the scripted projects, the scripted portion will be graded on a scale from F to B+ (presumably leaning toward the latter!). That is, if the project as stated is completed and done well, it will receive a B+. An extension of the project -- on of your own choosing -- is worth an additional full grade level. Each project will suggest possible theoretical or applications-based extensions. Computer vision is very much an experimental field. This piece of each project (and the final project as a whole) is meant to encourage you to experiment! No idea is too wild, though some may be too difficult. Lots of information is available from the links on the references page.

The final project will be worth the equivalent of two of the scripted projects.

Collaboration and the Honor Code