原理

face_recognition是最简单的人脸识别库，该模型的准确率为 99.38%
代码仅需要三行：

import face_recognition
image = face_recognition.load_image_file("your_file.jpg")
face_locations = face_recognition.face_locations(image)

一、效果预览

face_locations 返回图片上所有人脸的坐标位置，配合cv2库在图片上将所有坐标画出来。

二、实现过程

import face_recognition
import cv2
​
#加载图片
image = face_recognition.load_image_file("1.jpg")
#识别图片上所有人脸的位置
face_locations = face_recognition.face_locations(image)
​
img = cv2.imread("1.jpg", cv2.IMREAD_COLOR)
​
for i in face_locations:
    #face_locations 返回参数位置说明 (top, right, bottom, left)
    top = i[0]
    right = i[1]
    bottom = i[2]
    left = i[3]
​
    # param 1:图片
    # param 2:左上角坐标
    # param 3:右下角坐标
    # param 方框颜色
    # 方框大小
    cv2.rectangle(img, (left, top), (right, bottom), (255, 255, 255), 2)
    # 显示图片
    cv2.imshow("1.jpg", img)
    # 保存为新文件
    cv2.imwrite("new.jpg", img)
​
cv2.waitKey(0)
cv2.destroyAllWindows()