Question: image compression use face detection as heuristic

liuliu / ccv

C-based/Cached/Core Computer Vision Library, A Modern Computer Vision Library

http://libccv.org

Other

7.08k stars 1.72k forks source link

Question: image compression use face detection as heuristic #133

Closed morenoh149 closed 9 years ago

morenoh149 commented 9 years ago

Has anyone ever used face detection as a heuristic for image compression? How would one develop this?

meaning, detect regions of a photo with faces. Compress these regions less than the other parts of the photo.

liuliu commented 9 years ago

I don't see anything particular on this. Feel free to start a new GitHub project on this. The direction probably should be 1). Find faces in original photo; 2). Do compression as normal; 3). Do de-compression and then get the diff between original and compressed around faces; 4). Save the diff separately (do Huffman encoding all over again probably) into metadata.

Reverse the steps to recover.

morenoh149 commented 9 years ago

started it here https://github.com/morenoh149/FaceCompress :+1: