Face detection with Python + dlib

Installation

You need CMAKE, Boost, Boost Python, G ++ and more before installing dlib

Put OpenCV / scikit-image in the input / output of the image file This time I use OpenCV

code

You can get a facial score with dlib It feels like it will detect even if it is tilted a little, so even if you detect it while rotating it, the rotation angle may be about 45 degrees each. I screened with a somewhat low score (CUT_OFF value) and thought that it might be possible to adjust the angle so that the score is high while rotating the narrowed area, but it takes too much time The type that appears in the detection result seems to correspond to the type of detector (whether the face is front or sideways?) If you comment out, 68 feature points will be extracted from the detected face. If you know which points correspond to the eyes, nose, and contour, you can improve it a little more.

`detect_face.py`


# -*- coding: utf-8 -*-

from __future__ import absolute_import
from __future__ import division
from __future__ import print_function

import os, sys, imghdr, shutil, dlib, cv2

CWD = os.getcwd()
DIR_ORIGIN = CWD + '/images/'
DIR_DESTINATION = CWD + '/faces/'
CUT_OFF = -0.1

detector = dlib.get_frontal_face_detector()
# predictor = dlib.shape_predictor('./shape_predictor_68_face_landmarks.dat')

def getFaces(path_full):
    results = []
    image = cv2.imread(path_full)
    height, width = image.shape[:2]
    rects, scores, types = detector.run(image, 1, CUT_OFF)
    for i, rect in enumerate(rects):
        top, bottom, left, right = rect.top(), rect.bottom(), rect.left(), rect.right()
        if min(top, height - bottom - 1, left, width - right - 1) < 0:
            continue
        results.append({
            'image'       : image[top : bottom, left : right],
            'score'       : scores[i],
            'orientation' : types[i]
        })
        # shape = predictor(image, rect)
        # for i in range(shape.num_parts):
        #     print(shape.part(i))
    return results

count = 1
for path, subdirs, files in os.walk(DIR_ORIGIN):
    for name in files:
            path_full = os.path.join(path, name)
            if imghdr.what(path_full) in ['jpeg']:
                faces = getFaces(path_full)
                for face in faces:
                    file_name = '{destination_dir}/{score}_{type}_{count}_dlib.jpg'.format(
                        destination_dir = DIR_DESTINATION,
                        score = face['score'],
                        type = int(face['orientation']),
                        count = count
                    )
                    cv2.imwrite(file_name, face['image'], [cv2.IMWRITE_JPEG_QUALITY, 100])
                    count += 1
                print(path_full)

Reference site

Publish the know-how of creating a similar image search service for AV actresses by deep learning with chainer

Face detection program that links OpenCV and dlib

Machine learning library dlib

OpenFace 0.2.0: Higher accuracy and halved execution time

Summary of findings on face detection with OpenCV, Dlib and OpenFace

How to create a face image data set used in machine learning (3: Face image generation from candidate images 1)