CNS179/EE150/CS282 - Readings in visual recognition
Fall 2001 - California Institute of Technology
Organizational Meeting
2:00pm, Friday, October 5, 2001. Room 139, Moore
Building (#93)
Objective
Overview of the literature in visual recognition from a computational
perspective. The students attending the class will read and present the
most significant papers with the aim to understand:
-
taxonomy of recognition problems
-
main computational approaches
-
current open problems
Possible outcomes of the class are a paper surveying the literature in
a specific area, the implementation of published algorithms and variations
thereof, a psychophysical experiment, the collection of image databases
for testing algorithms and human subjects.
Overview
The class will meet for three hours every week. Each week a small number
of related papers will be presented and discussed by students attending
the class. Each student will have to present a subset of the papers in
the reading list. The final grade will depend on: bi-weekly homework,
final essay, presentations. Projects may be pursued for credit during the
winter quarter.
Syllabus
Week 1 (Fri, Oct 5): organizational meeting, overview of the reading
material, presentation assignment
Week 2 (Fri, Oct 12): Global image-based methods
Week 3 (Fri, Oct 19): Point-based matching & invariants
Week 4 (Fri, Oct 26): Neural Networks and Support Vector Machines (I)
Week 5 (Fri, Nov 2): Neural Networks and Support Vector Machines (II)
Week 6 (Fri, Nov 9): Parts and Structure (I)
Week 7 (Fri, Nov 16): Parts and Structure (II)
Week 8 (Fri, Nov 23): Overview of Psychophysics (alternative: Image
retrieval)
Week 9 (Fri, Nov 30): Overview of physiology and clinical studies (alternative:
Faces and biometrics, OR Aspect graphs)
Week 10 (Fri, Dec 7): Presentation and summary of course material
Contacts
Instructor: Pietro Perona
TA: Fei Fei Li (feifeili@vision.caltech.edu), 395-3695