R + face detection a good mix?

Previous Topic Next Topic
classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view

R + face detection a good mix?

Hello, a newbie here just trying to figure out where to start with R.

I want to build an algorithm that can detect 'duplicate' photos from a series of photos - the common example is when you take multiple photos of a group of people, hoping that one shot captured everyone smiling. Any batch of kid photos would be another example because they rarely sit still for the camera.

One approach is to use timestamps to find photos taken at about the same time, and match the position & spatial relationship of identical faces. I think OpenCV has a face recognition lib that is well-used.

Is R something I could use to build an algorithm matches faces as above with a specified degree of tolerance? (i.e. one person in the group blinked or moved)  

There has been a lot of news recently about sophisticated new image processing algos in Google+ photos, Facebook, etc for finding good photos and hiding bad ones.  Has something like this already been done?