1. import numpy as np
    2. import cv2
    3. import os
    4. from sklearn.model_selection import train_test_split
    7. def open_img(path_):
    8.     # 图片
    9.     img_list = []
    10.     # 各图片对应的人名
    11.     labels_ = []
    12.     # 仅人名(无重复)
    13.     label = []
    14.     for file in os.listdir(path_):
    15.         if file == '.anonr' or file == '.htaccess':
    16.             continue
    17.         label.append(file)
    18.         position = path_+'/'+file
    19.         for parent, dirnames, filenames in os.walk(position):
    20.             for filename in filenames:
    21.                 if filename == '.htaccess':
    22.                     continue
    23.                 labels_.append(len(label)-1)
    24.                 img = cv2.imread(position+'/'+filename, cv2.IMREAD_GRAYSCALE)
    25.                 img = cv2.resize(img, (250, 350))
    26.                 img_list.append(img)
    27.     return img_list, labels_
    30. def train_and_test(img_list, labels_):
    31.     x_train, x_test, y_train, y_test = train_test_split(img_list, np.array(labels_),
    32.                                                         test_size=0.3, random_state=42, stratify=np.array(labels))
    33.     recognizer = cv2.face.FisherFaceRecognizer_create()
    34.     recognizer.train(x_train, y_train)
    35.     right = 0
    36.     false = 0
    37.     for cnt in range(len(y_test)):
    38.         test_label = recognizer.predict(x_test[cnt])
    39.         if (test_label == y_test[cnt]).any():
    40.             right += 1
    41.         else:
    42.             false += 1
    43.     return right/(right+false)
    46. if __name__ == '__main__':
    47.     path = "../faces"
    48.     imgList, labels = open_img(path)
    49.     print('人脸识别的准确率为：', train_and_test(imgList, labels))