I am working on a medical project to identify a syndrome based on the dimensions of face as a whole, eye, nose and mouth.

Just to make it clear, I am not looking for facial recognition (to input a face and find the matching face). What I am looking for is to give a picture as input and get nose, eye face and mouth in the shape they are and not as rectangles.

I was looking for some libraries that could help me achieve this and I found OpenCV could do this. However what I found was it could only give me as rectangles.

Is there any library that could help me do this?

The library can be in Java or Python or Ruby or native libraries that can be used with one of these languages.

I dont want to identify like this:

Identifies face parts in rectangles

I want to identify like this:

Identifies face parts in realistic shapes


2 Answers 2


The closest I could find to do what I wanted was flandmark.



To recognize the face there are some libraries (e.g facerec) that could help you to achieve this and work with the Python language.


  • 2
    This again seems to help me detect rectangles. I went through the library and I couldn't find a way to get them in the shape they are. Am I missing something? Jan 13, 2015 at 18:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.