verifid / facereg Goto Github PK

facereg is a module for face recognition with OpenCV and Deep Learning.

For now it can be used for just images. It is easy to use with a handy feature which downloads images from Google for you with given keywords to create dataset/s.

Uses two different technics CNN and HoG for recognition based on dlib's face recognition system with using face_recognition. facereg has totally three different layers and only recognizer has connection on encoder.

• CMake
• All dependencies are listed on requirements.txt and will be installed when you install with pip.

• Install module using pip:

  $ pip install facereg
  

• Download the latest facereg library from: https://github.com/verifid/facereg and install module using pip:

  $ pip install -e .
  

• Extract the source distribution and run:

  $ python setup.py build
  $ python setup.py install
  

• google_images:

import os
from facereg import google_images

output_directory = os.getcwd() + '/datasets' # directory path where you want to save photos
image_paths = google_images.download('michael jordan', limit=3)

• face_encoder:

import os
from facereg import face_encoder

datasets_path = os.getcwd() + '/datasets'
encodings_path = os.path.dirname(os.path.realpath(__file__)) + '/encodings.pickle'
# these are default values for this method
face_encoder.encode_faces(datasets=datasets_path, encodings=encodings_path, detection_method='cnn')

• recognize_faces:

from facereg import recognize_faces

image_path = 'DIRECTORY PATH OF YOUR_IMAGE'
names = recognize_faces.recognize(image_path)
# returns found names from your datasets

• Download images

# -d: keyword, -l: limit
$ python -m facereg -d 'michael jordan'
$ python -m facereg -d 'michael jordan' -l 5

• Recognition

# -i: Directory path for image
$ python -m facereg -i tests/resources/michael_jordan.jpeg