6.897 Seminar on Vision-Based Object Recognition (H)

L. M2-5, room 34-301
Professor Trevor Darrell, trevor@csail.mit.edu
Prereq.: a previous vision course
3-0-6

This subject does not qualify as an engineering concentration subject.

This seminar will review contemporary literature in computer vision-based object recognition.  Topics well be determined by the interest of participants and may include locally invariant feature detectors and descriptors, local shape models, generative models for object representation, boosting-based classifiers, part-based recognition with conditional random fields and/or support vector machines, integration of multiple cues and modalities, use of lexical dictionaries and automatic data collection from the web, and innovative applications ranging from robot navigation to pervasive computing.