Part 1: Texture analysis-synthesis

Data: http://www.cis.upenn.edu/~jshi/Teaching/cis700-002/Project/TEXTURE/

You need to construct a program for:

a) given an texture image, synthesis a “similar” looking image of the same size;

b) given an texture image, “grow” the texture to be twice as large, by adding new elements to the original image;

c) given a pair of texture images, compute a similar score so that for each image, we can rank all other images by their similarity to it.

It is recommended you use Efros&Freeman algorithm for part a) and b), and use Portilla&Simoncelli, Robner-Tomasi-Guibas for part c).

Part 2: Face detection

Data: http://www.cis.upenn.edu/~jshi/Teaching/cis700-002/Project/FACE/

The training data for face and non-face are in “Image” subdirectory. The Matlab function for reading the images is in the “matlab” subdirectory, named “getfaces.m”.

You need to construct a program for:

a) Given an image patch, classifying it as face vs. nonface. The algorithm should use 75% of the faces, and non-faces in the database as training example, and report ROC curve on the remaining 25% in the database.

b) Given a full image, automatically detect the location and size of the faces in the image.

It is recommended to use Viola&Jones boosting algorithm for this part.

Part 3: Motion Tracking

Data: http://www.cs.cmu.edu/~jshi/Teaching/cis700-002/Project/TRACKING

You need to construct a program for:

a) Given an initial labeling of the person in the first frame, track the person as he walks around the room. Track as many frames as you can.

b) Same as part a) except to provide more detailed tracking of the head, upper body, and lower body.

It is recommended to use Isard&Blake Algorithm for this part.

Part 4: Image Segmentation

Data: http://www.cs.cmu.edu/~jshi/Teaching/cis700-002/Project/SEG

You need to construct a program to:

a) given an image, and a number “k”, segment the image into k elements;

b) Combine face detection to segment the human out of its background.