Integrate AI models

[cosmin42]

Add tensorflow / other libraries to the current project.

Example of high level usecases

1. Image enhancement: Given an image output another one which would be an improvement upon the original image.
2. Best LUT (a): Given an image and a set of LUTs find the one that first "best".
3. Best LUT (b): Given an image, get the LUT that would fit "best".
4. Best layout: Given a set of images and a set of layouts (TBD how the layouts can be fed to the program) find the "best" association between the images and the collages.
5. Face detection/segmentation: Given a set of images extract the human faces. This should result in a set of images containing the segments.
6. Object detection and segmentation: Given an image get the pieces from the image that are separate objects.
7. Image restoration: Given a damaged image (noisy image, image with missing parts, scratched etc) get an approximation of the original.

"best" - TBD what it means more precisely, to be researched what it means in the already existing products.