face_recognition是世界上最简洁的人脸识别库，你可以使用Python和命令行工具提取、识别、操作人脸。
face_recognition的人脸识别是基于业内领先的C 开源库 dlib中的深度学习模型，用Labeled Faces in the Wild人脸数据集进行测试，有高达99.38%的准确率。但对小孩和亚洲人脸的识别准确率尚待提升。
face_recognition可以产生很多有趣的应用。
官方原文代码见：https://github.com/ageitgey/face_recognition/blob/master/examples/face_recognition_knn.py
face_recognition包括三个部分的代码
1、训练数据集
2、预测数据集
3、输出标签
训练数据集要遵循一定的数据格式，可以是以文件夹名称为标识的一个个图片，也可以是以文件名为标识的单个图片，当然前者每个人的图片越多，训练越充分。
训练数据集首先要检测出人脸，多个或零个均非合法的人脸
然后将图片二进制传入X，图片标识传入y，进行训练
训练图片是使用sklearn的KNN近邻分类器（KNeighborsClassifier）进行训练的，这个近邻的个数可以调节。
训练完成后写入模型文件，写入模型文件的好处是一次训练，后续可以直接使用，毕竟训练的时间过长，而用户可以实时添加人脸，难点在于如何增量进行训练？还没想到好办法。
预测过程中最大的困惑是neighbors的返回值，以及对返回值的处理，尤其是distance，这个distance关系到预测的准确与否，无论如何knn都会返回最近的距离和标签，但这个标签正确与否就不知道了，所以阈值设置很重要，我这边设置的是0.5。
最后是在图片上标注出人脸矩形和识别出的人物标签。
我这边是用的ORL数据集，以及从网上找到刘德华、成龙和我的照片。
代码语言：javascript复制import math
from sklearn import neighbors
import os
import os.path
import pickle
from PIL import Image, ImageDraw
import face_recognition
from face_recognition.face_recognition_cli import image_files_in_folder
ALLOWED_EXTENSIONS = {'bmp','png', 'jpg', 'jpeg'}
# 对指定的训练图片文件夹进行训练
def train(train_dir, model_save_path=None, n_neighbors=None, knn_algo='ball_tree', verbose=False):
"""
Trains a k-nearest neighbors classifier for face recognition.
:param train_dir: directory that contains a sub-directory for each known person, with its name.
(View in source code to see train_dir example tree structure)
Structure:
<train_dir>/
├── <person1>/
│ ├── <somename1>.jpeg
│ ├── <somename2>.jpeg
│ ├── ...
├── <person2>/
│ ├── <somename1>.jpeg
│ └── <somename2>.jpeg
└── ...
:param model_save_path: (optional) path to save model on disk
:param n_neighbors: (optional) number of neighbors to weigh in classification. Chosen automatically if not specified
:param knn_algo: (optional) underlying data structure to support knn.default is ball_tree
:param verbose: verbosity of training
:return: returns knn classifier that was trained on the given data.
"""
X = []
y = []
# 循环获取训练集图片
for class_dir in os.listdir(train_dir):
if not os.path.isdir(os.path.join(train_dir, class_dir)):
continue
# 遍历当前任务的每一张图片
# image_files_in_folder，这个地方是获取文件夹下的所有图片文件，可以修改其中的图片类型
# 默认是jpg|jpeg|png，后来追加了bmp
print('training picture of {}'.format(class_dir))
for img_path in image_files_in_folder(os.path.join(train_dir, class_dir)):
# 加载图片文件，其实是numpy数组
image = face_recognition.load_image_file(img_path)
# 获取人脸检测框
face_bounding_boxes = face_recognition.face_locations(image)
# 多个人物或者0个人物不处理
if len(face_bounding_boxes) != 1:
# If there are no people (or too many people) in a training image, skip the image.
if verbose:
print("Image {} not suitable for training: {}".format(img_path, "Didn't find a face" if len(face_bounding_boxes) < 1 else "Found more than one face"))
else:
# Add face encoding for current image to the training set
X.append(face_recognition.face_encodings(image, known_face_locations=face_bounding_boxes)[0])
y.append(class_dir)
# 设置KNN分类器的近邻数
# Determine how many neighbors to use for weighting in the KNN classifier
if n_neighbors is None:
# n_neighbors = int(round(math.sqrt(len(X))))
n_neighbors = 3
if verbose:
print("Chose n_neighbors automatically:", n_neighbors)
# 创建KNN分类器，并进行训练
knn_clf = neighbors.KNeighborsClassifier(n_neighbors=n_neighbors, algorithm=knn_algo, weights='distance')
knn_clf.fit(X, y)
# 保存KNN训练结果
if model_save_path is not None:
with open(model_save_path, 'wb') as f:
pickle.dump(knn_clf, f)
return knn_clf
# 对指定的预测图片进行预测
def predict(X_img_path, knn_clf=None, model_path=None, distance_threshold=0.5):
"""
Recognizes faces in given image using a trained KNN classifier
:param X_img_path: path to image to be recognized
:param knn_clf: (optional) a knn classifier object. if not specified, model_save_path must be specified.
:param model_path: (optional) path to a pickled knn classifier. if not specified, model_save_path must be knn_clf.
:param distance_threshold: (optional) distance threshold for face classification. the larger it is, the more chance
of mis-classifying an unknown person as a known one.
:return: a list of names and face locations for the recognized faces in the image: [(name, bounding box), ...].
For faces of unrecognized persons, the name 'unknown' will be returned.
"""
# 校验当前文件类型
if not os.path.isfile(X_img_path) or os.path.splitext(X_img_path)[1][1:] not in ALLOWED_EXTENSIONS:
raise Exception("Invalid image path: {}".format(X_img_path))
# 校验当前模型文件和knn模型，两个不能全空
if knn_clf is None and model_path is None:
raise Exception("Must supply knn classifier either thourgh knn_clf or model_path")
# 加载训练好的KNN模型
if knn_clf is None:
with open(model_path, 'rb') as f:
knn_clf = pickle.load(f)
# 加载图片，获取人脸检测框
X_img = face_recognition.load_image_file(X_img_path)
X_face_locations = face_recognition.face_locations(X_img)
# print('predict {}'.format(X_img_path))
# 如果未找到人脸，返回[]
if len(X_face_locations) == 0:
return []
# 对测试图片进行编码转换，转换为numpy数组
faces_encodings = face_recognition.face_encodings(X_img, known_face_locations=X_face_locations)
# 通过KNN模型找到最佳匹配的人脸
closest_distances = knn_clf.kneighbors(faces_encodings, n_neighbors=3)
# # 返回值indices：第0列元素为参考点的索引，后面是(n_neighbors - 1)个与之最近的点的索引
# # 返回值distances：第0列元素为与自身的距离(为0)，后面是(n_neighbors - 1)个与之最近的点与参考点的距离
# closest_distances= [[0.34997745 0.3750366 0.37819395]]
# closest_distances= [[ 5 12 11]]
# for i in closest_distances:
# print('closest_distances=',i)
are_matches = []
# are_matches = [closest_distances[0][i][0] <= distance_threshold for i in range(len(X_face_locations))]
for i in range(len(X_face_locations)):
are_matches.append( closest_distances[0][i][0] <= distance_threshold)
#print('predict value=', closest_distances[0][i][0])
# print('knn_clf.predict(faces_encodings)=',knn_clf.predict(faces_encodings))
# 预测分类
return [(pred, loc) if rec else ("unknown", loc) for pred, loc, rec in zip(knn_clf.predict(faces_encodings), X_face_locations, are_matches)]
# 在图片上输出预测标签
def show_prediction_labels_on_image(img_path, predictions):
"""
Shows the face recognition results visually.
:param img_path: path to image to be recognized
:param predictions: results of the predict function
:return:
"""
# 打开图片，获取句柄
pil_image = Image.open(img_path).convert("RGB")
draw = ImageDraw.Draw(pil_image)
# 对预测结果进行遍历
for name, (top, right, bottom, left) in predictions:
# 在人脸周边画矩形框
draw.rectangle(((left, top), (right, bottom)), outline=(0, 0, 255))
# There's a bug in Pillow where it blows up with non-UTF-8 text
# when using the default bitmap font
name = name.encode("UTF-8")
# 在人脸地图输出标签
text_width, text_height = draw.textsize(name)
draw.rectangle(((left, bottom - text_height - 10), (right, bottom)), fill=(0, 0, 255), outline=(0, 0, 255))
draw.text((left 6, bottom - text_height - 5), name, fill=(255, 255, 255, 255))
# Remove the drawing library from memory as per the Pillow docs
del draw
# Display the resulting image
pil_image.show()
if __name__ == "__main__":
# 指定训练集路径和预测集路径
train_dir=r'C:PythonPycharmdocxprocesspictureORL'
test_dir=r'C:PythonPycharmdocxprocesspicturepredict'
# 第一步，通过KNN分类器进行训练，并存储模型文件
print("Training KNN classifier...")
classifier = train(train_dir, model_save_path="trained_knn_model.clf", n_neighbors=3)
print("Training complete!")
# 第二步，使用训练分类器，对未知图片进行预测
for image_file in os.listdir(test_dir):
full_file_path = os.path.join(test_dir, image_file)
print("Looking for faces in {}".format(image_file))
# 使用训练模型进行预测
predictions = predict(full_file_path, model_path="trained_knn_model.clf")
# 输出结果
for name, (top, right, bottom, left) in predictions:
print("- Found {} at ({}, {})".format(name, left, top))
# 图片上显示输出结果
show_prediction_labels_on_image(os.path.join(test_dir, image_file), predictions)人物1的识别，分类准确，距离为0
人物2的识别，分类准确，距离为0
合照的识别，里面包括周星驰、刘德华、李连杰、成龙，但周星驰和李连杰不在训练集内，周星驰未识别出来，刘德华和成龙识别正确，但李连杰识别出的最近距离甚至比刘德华本人还接近。
刘德华本人的识别，分类准确，距离为0.35左右
刘亦菲的识别，分类准确，距离为0.68左右
本人的识别，分类准确，距离为0.42左右，我只有两张训练照片，一张是身份证，一张是正照。
