V22.0480-001, Computer Vision

Davi Geiger

Mon/Wed 9:30 a.m.-10:45 a.m.

CIWW 109 (Moved to the Multimedia Center, 719 Broadway, 1221, 12 floor)


Course Description

This course will look at basic ideas on Computer Vision and how they work. Students will be encouraged to understand well the material as well as develop software that works on images. The course will cover basic Image Processing tools (convolution, filtering, and multiscale representation of images), Edge detection, Contour Detection/Segmentation, Character Recognition, Skeletonization, Stereo Vision, Motion (Optial Flow and 3D Structure), Shape From Shading and Color, Recognition.

Reading Material

Instead of books I recommend to read suplement course material from http://www.dai.ed.ac.uk/CVonline and some specific materials are linked below.

Evaluation Process

There will be homeworks, about every two weeks, with programming assignments and written assignments. The student will be exposed to both, experiments (programming) and theory (written homeworks).
There will be a course project, starting at the beginning of the course (by the fourth lecture it starts). There will be a choice of different projects. The project will "substitute the mid term exam". There will be a final exam for the students to fully review the material.


Class 1: Introduction

Class 2: Surface Reflectance, Radiance and Irradiance.

Class 3: Filtering and Convolution http://www.dai.ed.ac.uk/CVonline/LOCAL_COPIES/MARSHALL/node15.html

Class 4: Edge Enhacement and Detection. Problem set 1. 2 pages postscript .

Class 5: Curvature Enhacement and Detection. http://www.dai.ed.ac.uk/CVonline/LOCAL_COPIES/MARBLE/medium/surfaces/surfaces.htm .

Class 6 and 7: Learning with perceptron.

Class 8: Multilayer Networks.

PROJECT: Here you find a description of the project , projectdata.zip is a zip file with testing and training data for numerals. See the readme file . Jong Oh have offered a sample of a Neural Net code to add numbers with one hidden layer: Neural Net Sample Code

Class 9 and 10: Stereo Vision.

Class 11, 12, and 13: Motion. Optical Flow and Structure from Motion. homework 2

Class 14 and 15: Detecting Snakes and the Shortest Path Algorithm.

Class 15 and 16: Tutoring on Optical Flow.

Class 17 and 18: Texture.