COMP 4060/7950: Course project

The course project is probably the most important part of this course. The goal is to give you some flavor on how to do research. You can do the project individually, or as a group of two people. Group projects should be more substantial than individual ones. I generally do not recommend group sizes of more than two, since large groups typically create more overhead. But if you have a well-thought project that justifies larger groups, please discuss with me.

Computer vision is a very broard area, so it is impossible to cover every aspect of it in the lectures. Depending on the specific projects, you might need to do some reading ahead of the course schedule and learn materials not covered in the classroom. I can provide pointers and help if needed.

IMPORTANT: for students who are also taking the image processing course from Prof.Neil Bruce, you are allowed to use the same project for both courses, if the following conditions are met:
  • You should inform Prof.Bruce and me that you will use the same project for both courses.
  • Your project should be a good fit for both courses (e.g. a project on object recognition is perfectly fine for this course, but is unlikely to be a suitable project for Prof.Bruce's course unless you do something interesting on the image processing side). You are welcome to discuss with both Prof.Bruce and myself. But eventually it is your responsibility to find a suitable project.
  • You still need to meet all the deadlines and requirements of either course separately.
  • We will expect your project to be more substantial.
IMPORTANT: you are encouraged to choose a course project related to your own research and reuse the resources (e.g. code, dataset, etc) you already have from your research. However, I expect you to do something new during the semester. You are not allowed to submit any work you did before taking this course as your course project.
There are three deliverables for the course project. More details on these deliverables and how to submit them will be announced in due course.

Proposal

I suggest you discuss your project with me and get my approval as early as possible. By Nov 5 (I should have already approved your project by this date), each individual/group should send me a one-page proposal. It should contain the following information:
  • Name(s) of people in the group
  • Problem to be addressed
  • Techniques to be used or developed
  • Datasets to be used or collected
  • Possible issues forseen
  • Approximate timeline for the project to be completed
  • Important references (including existing codes to be used)

Final report

The final project report is due on Dec 10 (extended to Dec 15). It should be written in CVPR style (Latex/Word template). The report should be 4-6 pages in this format.

Project presentation

Each individual/group will also be required to give a presentation of their project. The details will be announced in due course.

Making progress

One difficult aspect of doing research is that you need to have motivations to make progress. The proposal and final report due dates can help motivate you to finish certain things by certain dates, but there are no other checkpoints in between. To help you stay motivated, I will be happy to schedule regular meetings with you and your group, if you think that will help you stay on track.

Conference and workshop submission

If you come up with good results in your course project, you are encouraged to submit them (possibly with some polishing) as a paper to a conference or workshop. Some possible choices are (more to be added): If you want more information, you can check out the conference calendar for computer vision, image analysis and related topics. If you are interested in submitting your project to any of these or other conferences/workshops, you should let both your advisor and me know as soon as possible, so we can help you plan your time accordingly.