face_recognition/face_recognition/api.py

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

import PIL.Image
import dlib
import numpy as np
from PIL import ImageFile

try:
    import face_recognition_models
except Exception:
    print("Please install `face_recognition_models` with this command before using `face_recognition`:\n")
    print("pip install git+https://github.com/ageitgey/face_recognition_models")
    quit()

ImageFile.LOAD_TRUNCATED_IMAGES = True

face_detector = dlib.get_frontal_face_detector()

predictor_68_point_model = face_recognition_models.pose_predictor_model_location()
pose_predictor_68_point = dlib.shape_predictor(predictor_68_point_model)

predictor_5_point_model = face_recognition_models.pose_predictor_five_point_model_location()
pose_predictor_5_point = dlib.shape_predictor(predictor_5_point_model)

cnn_face_detection_model = face_recognition_models.cnn_face_detector_model_location()
cnn_face_detector = dlib.cnn_face_detection_model_v1(cnn_face_detection_model)

face_recognition_model = face_recognition_models.face_recognition_model_location()
face_encoder = dlib.face_recognition_model_v1(face_recognition_model)


def _rect_to_css(rect):
    """
    Convert a dlib 'rect' object to a plain tuple in (top, right, bottom, left) order

    :param rect: a dlib 'rect' object
    :return: a plain tuple representation of the rect in (top, right, bottom, left) order
    """
    return rect.top(), rect.right(), rect.bottom(), rect.left()


def _css_to_rect(css):
    """
    Convert a tuple in (top, right, bottom, left) order to a dlib `rect` object

    :param css:  plain tuple representation of the rect in (top, right, bottom, left) order
    :return: a dlib `rect` object
    """
    return dlib.rectangle(css[3], css[0], css[1], css[2])


def _trim_css_to_bounds(css, image_shape):
    """
    Make sure a tuple in (top, right, bottom, left) order is within the bounds of the image.

    :param css:  plain tuple representation of the rect in (top, right, bottom, left) order
    :param image_shape: numpy shape of the image array
    :return: a trimmed plain tuple representation of the rect in (top, right, bottom, left) order
    """
    return max(css[0], 0), min(css[1], image_shape[1]), min(css[2], image_shape[0]), max(css[3], 0)


def face_distance(face_encodings, face_to_compare):
    """
    Given a list of face encodings, compare them to a known face encoding and get a euclidean distance
    for each comparison face. The distance tells you how similar the faces are.

    :param faces: List of face encodings to compare
    :param face_to_compare: A face encoding to compare against
    :return: A numpy ndarray with the distance for each face in the same order as the 'faces' array
    """
    if len(face_encodings) == 0:
        return np.empty((0))

    return np.linalg.norm(face_encodings - face_to_compare, axis=1)


def load_image_file(file, mode='RGB'):
    """
    Loads an image file (.jpg, .png, etc) into a numpy array

    :param file: image file name or file object to load
    :param mode: format to convert the image to. Only 'RGB' (8-bit RGB, 3 channels) and 'L' (black and white) are supported.
    :return: image contents as numpy array
    """
    im = PIL.Image.open(file)
    if mode:
        im = im.convert(mode)
    return np.array(im)


def _raw_face_locations(img, number_of_times_to_upsample=1, model="hog"):
    """
    Returns an array of bounding boxes of human faces in a image

    :param img: An image (as a numpy array)
    :param number_of_times_to_upsample: How many times to upsample the image looking for faces. Higher numbers find smaller faces.
    :param model: Which face detection model to use. "hog" is less accurate but faster on CPUs. "cnn" is a more accurate
                  deep-learning model which is GPU/CUDA accelerated (if available). The default is "hog".
    :return: A list of dlib 'rect' objects of found face locations
    """
    if model == "cnn":
        return cnn_face_detector(img, number_of_times_to_upsample)
    else:
        return face_detector(img, number_of_times_to_upsample)


def face_locations(img, number_of_times_to_upsample=1, model="hog"):
    """
    Returns an array of bounding boxes of human faces in a image

    :param img: An image (as a numpy array)
    :param number_of_times_to_upsample: How many times to upsample the image looking for faces. Higher numbers find smaller faces.
    :param model: Which face detection model to use. "hog" is less accurate but faster on CPUs. "cnn" is a more accurate
                  deep-learning model which is GPU/CUDA accelerated (if available). The default is "hog".
    :return: A list of tuples of found face locations in css (top, right, bottom, left) order
    """
    if model == "cnn":
        return [_trim_css_to_bounds(_rect_to_css(face.rect), img.shape) for face in _raw_face_locations(img, number_of_times_to_upsample, "cnn")]
    else:
        return [_trim_css_to_bounds(_rect_to_css(face), img.shape) for face in _raw_face_locations(img, number_of_times_to_upsample, model)]


def _raw_face_locations_batched(images, number_of_times_to_upsample=1, batch_size=128):
    """
    Returns an 2d array of dlib rects of human faces in a image using the cnn face detector

    :param img: A list of images (each as a numpy array)
    :param number_of_times_to_upsample: How many times to upsample the image looking for faces. Higher numbers find smaller faces.
    :return: A list of dlib 'rect' objects of found face locations
    """
    return cnn_face_detector(images, number_of_times_to_upsample, batch_size=batch_size)


def batch_face_locations(images, number_of_times_to_upsample=1, batch_size=128):
    """
    Returns an 2d array of bounding boxes of human faces in a image using the cnn face detector
    If you are using a GPU, this can give you much faster results since the GPU
    can process batches of images at once. If you aren't using a GPU, you don't need this function.

    :param images: A list of images (each as a numpy array)
    :param number_of_times_to_upsample: How many times to upsample the image looking for faces. Higher numbers find smaller faces.
    :param batch_size: How many images to include in each GPU processing batch.
    :return: A list of tuples of found face locations in css (top, right, bottom, left) order
    """
    def convert_cnn_detections_to_css(detections):
        return [_trim_css_to_bounds(_rect_to_css(face.rect), images[0].shape) for face in detections]

    raw_detections_batched = _raw_face_locations_batched(images, number_of_times_to_upsample, batch_size)

    return list(map(convert_cnn_detections_to_css, raw_detections_batched))


def _raw_face_landmarks(face_image, face_locations=None, model="large"):
    if face_locations is None:
        face_locations = _raw_face_locations(face_image)
    else:
        face_locations = [_css_to_rect(face_location) for face_location in face_locations]

    pose_predictor = pose_predictor_68_point

    if model == "small":
        pose_predictor = pose_predictor_5_point

    return [pose_predictor(face_image, face_location) for face_location in face_locations]


def face_landmarks(face_image, face_locations=None, model="large"):
    """
    Given an image, returns a dict of face feature locations (eyes, nose, etc) for each face in the image

    :param face_image: image to search
    :param face_locations: Optionally provide a list of face locations to check.
    :param model: Optional - which model to use. "large" (default) or "small" which only returns 5 points but is faster.
    :return: A list of dicts of face feature locations (eyes, nose, etc)
    """
    landmarks = _raw_face_landmarks(face_image, face_locations, model)
    landmarks_as_tuples = [[(p.x, p.y) for p in landmark.parts()] for landmark in landmarks]

    # For a definition of each point index, see https://cdn-images-1.medium.com/max/1600/1*AbEg31EgkbXSQehuNJBlWg.png
    if model == 'large':
        return [{
            "chin": points[0:17],
            "left_eyebrow": points[17:22],
            "right_eyebrow": points[22:27],
            "nose_bridge": points[27:31],
            "nose_tip": points[31:36],
            "left_eye": points[36:42],
            "right_eye": points[42:48],
            "top_lip": points[48:55] + [points[64]] + [points[63]] + [points[62]] + [points[61]] + [points[60]],
            "bottom_lip": points[54:60] + [points[48]] + [points[60]] + [points[67]] + [points[66]] + [points[65]] + [points[64]]
        } for points in landmarks_as_tuples]
    elif model == 'small':
        return [{
            "nose_tip": [points[4]],
            "left_eye": points[2:4],
            "right_eye": points[0:2],
        } for points in landmarks_as_tuples]
    else:
        raise ValueError("Invalid landmarks model type. Supported models are ['small', 'large'].")


def face_encodings(face_image, known_face_locations=None, num_jitters=1, model="small"):
    """
    Given an image, return the 128-dimension face encoding for each face in the image.

    :param face_image: The image that contains one or more faces
    :param known_face_locations: Optional - the bounding boxes of each face if you already know them.
    :param num_jitters: How many times to re-sample the face when calculating encoding. Higher is more accurate, but slower (i.e. 100 is 100x slower)
    :param model: Optional - which model to use. "large" (default) or "small" which only returns 5 points but is faster.
    :return: A list of 128-dimensional face encodings (one for each face in the image)
    """
    raw_landmarks = _raw_face_landmarks(face_image, known_face_locations, model)
    return [np.array(face_encoder.compute_face_descriptor(face_image, raw_landmark_set, num_jitters)) for raw_landmark_set in raw_landmarks]


def compare_faces(known_face_encodings, face_encoding_to_check, tolerance=0.6):
    """
    Compare a list of face encodings against a candidate encoding to see if they match.

    :param known_face_encodings: A list of known face encodings
    :param face_encoding_to_check: A single face encoding to compare against the list
    :param tolerance: How much distance between faces to consider it a match. Lower is more strict. 0.6 is typical best performance.
    :return: A list of True/False values indicating which known_face_encodings match the face encoding to check
    """
    return list(face_distance(known_face_encodings, face_encoding_to_check) <= tolerance)