Face detection based on opencv

Face detection is a prerequisite for processing face pictures,Study if needed,Then the sampling of the data set can be combined with the face detection function for linked collection and filtering.此处,我们主要采用pythonProgramming language to process the face information in the picture.

First we need to install in our computeropencv,If there is a problem with the installation,可以尝试通过anacondaDo a new environment installation.

First we import the module

import cv2                          #opencv2库
import matplotlib.pyplot as plt     #plt绘图库
import numpy as np
import os
import scipy.io

在这里,We don't just have to detect the faces we collect,We also need to label the different faces that appear in a frame,If two faces are the same,Then we assume they are the same person,store their information together.(This part of the code is not released yet)

Get face information

in the code for face detection,We first need to load our cascaded classifier,在opencv中,The function of our cascaded classifier is ：CascadeClassifier
We load open sourcehaarcascade_frontalface_default.xml人脸分类器

下载链接：https://github.com/kipr/opencv/blob/master/data/haarcascades/haarcascade_frontalface_default.xml

We filter the information classified by our cascade classifier through the cascade classifier,Get the face data we want.我们可以使用 分类器的detectMultiScale方法.
该方法中的imageThe parameter needs to be an image information,scanleFactorParameters are parameters that specify how much the image size is reduced at each image scale,minNeighborsThe parameter specifies how many neighbors each candidate rectangle should retain,我们在这里对scanleFactor设置为1.5,对minNeighbors设置为5.

那么我们的代码如下：

face_cascade = cv2.CascadeClassifier('haarcascade_frontalface_default.xml')
faces = face_cascade.detectMultiScale(image = img, scaleFactor = 1.5, minNeighbors = 5)

我们得到的faces是一个向量数组（类似于pytorch中的张量,也就是tensor）,But its essence is an array,我们通过len方法获得faces数组中的个数,那么如果len(faces)<0,There are no faces detected in our images.

人脸裁剪

通过for循环遍历faces数组,Get our imagetensor实际上是numpy.Array的数据类型,并通过start_pos和end_posGet the start coordinate and the end coordinate on the diagonal in the 2D coordinate system.got a rectangle.

for (x, y, w, h) in faces:
        start_point = (x,y)
        end_point = (x+w,y+h)

        #image drawing
        cv2.rectangle(img, start_point, end_point, (255, 0, 0), 2)        #color参数的顺序是BGR,因此（255,0,0）Shown are blue lines,##The last parameter is the thickness of the rectangle.
        #图像裁剪,并保存为temp.png
        crop = img[y:(y + h), x:(x + w)]

人脸绘制

在裁剪后,We do the final processing on the image,Temporarily our processing results

finalimg = cv2.cvtColor(img, cv2.COLOR_RGB2BGR)
plt.figure(figsize=(5,5))
plt.imshow(finalimg) 
plt.axis("off")
plt.show()

The complete code is not released yet.

效果展示

输入图像

输出图像

而在实际的使用过程中,We put the same faces in the same folder,and ready for subsequent operations.

例子：

For similar picture streams,We may be able to use some technical means to do some future-oriented things～！