Compute the first five eigenfaces for both databases
Implement the Eigenface and Fisherface method...
Use a cross-validation scheme of leave-one-"expression or lighting" for the Yace Face Database. What expression/lighting is the worst in terms of accuracy?
Use a ten-fold cross-validation scheme and report the mean an stand deviation accuracies for the ORL database. Is there a difference statisical significance between the reported values?
Provide a runtime analysis for training and testing?
Evaluation
Send by email the generated codes until the begining of the next theorectical class