Simple Image Recognition

Simple Image Recognition system in C using Power Iteration to compute top K eigenfaces, alongwith some Python scripts to use the program

Training

1. Gather a labelled images dataset (name_i.png where name is the label, i the index of the specimen in the dataset)
2. Compute the mean face of the whole dataset, and store it
3. Center the dataset by substracting the mean face
4. Compute the first K eigenvectors of the dataset
5. Derive the eigenfaces, and the projected faces
6. Store mean face, eigenfaces, and projected faces

Classification

1. Gather dataset to classify
2. Substract the mean face from the images
3. Project the images on the eigenfaces
4. Find the closest labelled image using euclidian distance with a tolerance.
5. Output the label if found or warn if the tolerance can't be satisfied