Average Face:

eigenFaces:

cropped elf.tga:

No digital pic of myself, so i used obama
origional image:
cropped image:

group nonsmiling image IMG_0003:

group smiling image IMG_0006:

non class group image:

THE DREADED IMG_0002:

Overview of results:
Although I implemented the recomended "multiply the MSE by the distance of the face from the face mean and then dividing by the variance of the face", it the face detector still seems to prefer to pick regions of solid color...sometimes.
For the first 2 images i used a min_scale=.45 max_scale=.55 and scale=.01
The hard part for the next 2 was that since the faces themselves were smaller, but the images were the same size, I needed to use a scale of .04, min_scale=.60 and max_scale=.75

Unfortunately, due to massive time constraints, I was not able to fully complete the writeup and perfect my code.