Face Detector Pool

• Implement common face detection algorithms.

Requirements

• Python 3.6
• OpenCV 3
• Dlib
• Tensorflow 1.4

Current methods

• Dlib
• MTCNN

Usage

• Use scripts

 python run_face_analyze.py [--detector [method]] 
                            [--input [image file or folder]]                      
                            [--viz [visualize or not]]

For example, python run_face_analyze.py --detector dlib --input Yuniko --viz True

• Use high level class

  • Create FaceAnalyzer() instance with a detector.

  • Run full_analyze() will return detected faces and related facial landmarks.

• Use detector api

  • detect_faces() return face bounding boxes.
  • detect_facial_landmarks() return facial landmarks.

Outputs

• Detected face will be cropped and saved into output folder.
• Related landmarks will be saved into face_metadata.json with <key, value> = <img_path, landmarks>.