Dimensionality Reduction and face classification using PCA on AT&T Face dataset.

This is a numpy based implementation of PCA for dimensionality reduction and image classification using K-Nearest-Neighbors.

• To run the program in default mode run:

python3 main.py

• To run in PCA mode(default mode):

python3 main.py --mode=pca

• To run in image reconstruction mode:

python3 main.py --mode=image_recons --nPC=30(your choice)

It will also save few reconstructed image samples.

• To run in image classification mode:

python3 main.py --mode=face_classification --nPC=30(your choice)

With k=40, I am getting face recognition of 92.1 % on AT&T Face dataset.

Please check other flags for more information.