faze
A python module to analyze faces, provides functionalities like: face detection, landmark detection, pose estimation etc.
Requirements
- OpenCV ( 3.0.0+ )
- scikit-learn ( scikit-learn>=0.15.1)
- Dlib ( 18.16+ )
- You should also download the shape_predictor_68_face_landmarks.dat from here.
- Numpy ( numpy>=1.8.0 )
Structure
- detect: Face/landmark detection, face pose estimation, frontal transform of faces.
- recog: A submodule which caters to all the recognition methods. Currently, the following are supported.
- emotion
Getting started
Clone this repository to get started.
$ git clone https://github.com/Aravind-Suresh/faze
$ cd faze
To start using the functionalities, import and use. For example,
# imports necessary modules and methods
import faze
::
# dets stores the ROIs of faces detected
dets = faze.detect.faces(img)
# ROI is the first face detected
roi = dets[0]
# landmarks holds the facial landmarks of the face ROI
landmarks = faze.detect.landmarks(img, roi, "/path/to/shape_predictor")
How to contribute
If you think of any improvements to the project, feel free to make a pull request.