COMP-766: Shape Analysis in Computer Vision

Timings: T/Th: 13:00pm - 14:30pm
Room: McConnell 320
Instructor: Kaleem Siddiqi
Office: MC-420
Office Hrs: Tuesdays: 15:00pm - 16:00pm, or by appointment
Phone: 398-3371

Teaching Assistant: TBA

Course description

Object shape lies at the interface between vision and cognition, yet, truly general purpose theories of shape for applications in industry, bio-medicine, and robotics, have been notoriously difficult to formulate. In this research seminar we shall attempt to articulate the critical aspects of such a theory. The course will emphasize the interdisciplinary nature of the problem, drawing on insights from diverse areas ranging from psychology to singularity theory and classical mechanics. Topics to be covered include (but are not limited to): early vision, curve inference and fragment grouping, variational methods, geometric scale spaces, curve and surface evolution, level set techniques, shape segmentation, shape matching, object recognition.

Course format

The course will be organized around lectures as well as discussions of journal arcticles and papers which will be distributed on a weekly basis. Students will be expected to prepare short (2 page) critical summaries of selected articles and to participate actively in class. Whereas there is no assigned textbook, relevant reference material will be placed on reserve in the physical sciences and engineering library. Some background in differential equations, differential geometry and linear algebra will be helpful. A significant aspect of this course will be an independent research project, carried out in consultation with the instructor, on which each student will be expected to make a presentation at the end of the term.

Calendar

Assignments

  • Assignment 1 (PDF).
  • Assignment 2 (PDF).

    • Downloads

  • Trace Inference (PS) (PDF)
  • Relaxation Labeling (PS) (PDF)
  • Area/Length Flows (PS) (PDF)
  • Romeny's Front End Vision Book
  • Scale Space Notes (PDF)

    • News

  • Course packs should be available (for order) in the McGill bookstore on McTavish street, starting late this week or early next week. The course pack is the same from the one from two years ago, so you could in fact borrow it from a past student as well.
  • The calendar page is more or less complete, although dates and detailed content are subject to change. You can take a look at the many articles that are candidates for summaries, so as to plan in advance. Note that several of these articles come at the end of the course, once we have covered the necessary ground.
  • I have activated the discussion board under mycourses. The notes on curves from DoCarmo are also available there, under "content".

    • Course evaluation

  • Research project: 40%
  • Midterm examination: 25%
  • Assignments: 20%
  • Critical summaries: 15%
    • There will be two assignments worth 10% each. In addition you will be required to write critical summaries (no more than two single-spaced pages each) of 5 articles of your choice from the set of readings posted under the calendar. Each critique will comprise 3% of your total mark.

    All assignments and exams are to be carried out INDIVIDUALLY. In accordance with McGill University's Charter of Students Rights, students in this course have the right to submit in English or in French any written work that is to be graded. McGill University values academic integrity. Therefore all students must understand the meaning and consequences of cheating, plagiarism and other academic offences under the Code of Student Conduct and Disciplinary Procedures (see www.mcgill.ca/integrity for more information).

    Past Research Projects

    Below is a set of links to some of the research projects completed by students who took this course in the past. I am grateful to them for making this material available. 2002-2003 Winter Term
  • Maxime Descoteaux: Affine and Euclidean Geometric Heat Equation for Anisotropic Smoothing (HTML)/ (PDF)/ Experimental results
  • Simon Lacoste-Julien: Geometrical Analysis of an Area Minimization Flow (HTML)/ (PDF)
  • Cathy Laporte: Context-based aspect graphs for active object recognition (HTML)
  • Sandra Skaff: Perceptual illusions arising from dot displays (PDF).
    • 2003-2004 Fall Term
  • Shiyan Hu: Stochastic Completion Field With Probabilistic Transition (PDF)
  • Tran-Quan Luong: An Exploration of Stochastic Completion Fields (PDF)
  • Scott McClosskey: Shape From Pictures (PDF)
  • Frank Riggi: Matching Shapes by Probabilistic Graph Matching (PDF)
    • 2009-2010 Winter Term
  • Mohannad Elhamod: Shape From Shading Using Level Sets (PDF)
  • Chatavut Viriyasuthee: Learning Compatibility Coefficients (PDF)
  • Mostafa Abd El Meguid: Symmetries of Objects (PDF)

    • Links

  • All IEEE publications are available at IEEE EXPLORE .
  • Many science articles are available at JSTOR .
  • International Journal of Computer Vision Home Page .
  • McGill Libraries . Click on Muse and then on Journal Titles to do a search.
  • Elsevier Science Journals .
  • Classics in the History of Psychology .
  • Computer Vision Home Page.