After a really long time I come back to my blog. For some time recently I have been thinking about comics and Computer Vision. Comics and more generally Non-Photorealistic Rendering techniques http://isgwww.cs.uni-magdeburg.de/~stefans/npr/nprpapers.html
provide a different representation/sketching of visual scenes that might be useful in many cases.
Some random references include:
1. Talk by Aaron Hertzmann: Art, Vision, and Probability, given at the Symposium on Computational Photography and Video
2. For an introduction to issues of perception in art, two books: Understanding Comics, by Scott McCloud, and Picture This by Molly Bang
3. About Random Computer-Generated Comics, by John Pound
http://www.poundart.com/art/randcomix/about.html
The CMU face databases can be accessed from
http://www.ri.cmu.edu/projects/project_419.html
A nice collection of pointers to various CV datasets:
Databases you can use for benchmarking