(Sorry, the interactive parts below are not enabled)
Five classification methods are demonstrated. The methods, all based on
the eigenface representation, differ in the distance measure for
classification. There are over 1000 face images from six databases:
Cambridge, Bern, Yale, MIT, Harvard and our own, subject to varying viewing angles.
lighting directions,
with or without glasses,
and
expressions,
races and gender.
See also related papers.
Retrieve audio sounds according to the sound properties of the query
sounds, from a
database containing 16 classes of 409 sounds. You can compare among
various representation and search methods. Two search engines are
provided: an icon-based
search engine and a
list based search engine.
You can find a full report of experiments
here.
Retrieval of the Brodatz textures. You can compare among various
representation and search methods.
Image retrieval by color and/or texture. You can compare among various
representation and search methods.
Image retrieval using Relevance Feedback. You can compare among various
representation and search methods.
The On-Line MRF Sampler
generates a texture image according to an auto model specified by the
parameters provided by the user.
Refer to Markov Random Field Modeling in
Computer Vision for more details of the model.