Opencv + clion face recognition + face model training

OpenCV windows Version compilation +CLion Project import reference ：

CLion+OpenCV Identification ID number --- Identification card number _xxwbwm The blog of -CSDN Blog

Code ：

#include <iostream>
#include <opencv2/opencv.hpp>

using namespace cv;
using namespace std;

class CascadeDetectorAdapter : public DetectionBasedTracker::IDetector {
public:
    CascadeDetectorAdapter(cv::Ptr<cv::CascadeClassifier> detector) :
            IDetector(),
            Detector(detector) {
        CV_Assert(detector);
    }

    void detect(const cv::Mat &Image, std::vector<cv::Rect> &objects) {
        Detector->detectMultiScale(Image, objects, scaleFactor, minNeighbours, 0, minObjSize, maxObjSize);
    }

    ~CascadeDetectorAdapter() {

    }

private:
    CascadeDetectorAdapter();

    cv::Ptr<cv::CascadeClassifier> Detector;
};

DetectionBasedTracker *getTracker() {
    // OpenCV Built in model location F:\opencvWin\opencv\build\etc\lbpcascades
    String path = "F:\\opencvWin\\opencv\\build\\etc\\lbpcascades\\lbpcascade_frontalface.xml";
//    String path = "F:\\opencvWin\\facetrain\\samples\\data\\cascade.xml";// Can't recognize 

    Ptr<CascadeClassifier> classifier = makePtr<CascadeClassifier>(path);
    // Adapter 
    Ptr<CascadeDetectorAdapter> mainDetector = makePtr<CascadeDetectorAdapter>(classifier);

    Ptr<CascadeClassifier> classifier1 = makePtr<CascadeClassifier>(path);
    // Adapter 
    Ptr<CascadeDetectorAdapter> trackingDetector = makePtr<CascadeDetectorAdapter>(classifier1);
    // tracker 
    DetectionBasedTracker::Parameters DetectorParams;
    DetectionBasedTracker *tracker = new DetectionBasedTracker(mainDetector, trackingDetector, DetectorParams);
    return tracker;
}

int main() {
    DetectionBasedTracker *tracker = getTracker();

    // Turn on the tracker 
    tracker->run();
    // Get camera data 
    VideoCapture capture(0);

    Mat Sourceimg;
    Mat gray;
    Mat test;
    while (1) {
        capture >> Sourceimg;
        // Gray image processing 
        cvtColor(Sourceimg, gray, COLOR_BGR2GRAY);
        // Enhance contrast  ( Histogram equalization )
        equalizeHist(gray, gray);
        // Create and save a vector set of detected faces 
        std::vector<Rect> faces;
        // Gray image recognition processing 
        tracker->process(gray);
        // To get the results 
        tracker->getObjects(faces);

        for (Rect face : faces) {
            // Assign separately  bgra
            if (face.x < 0 || face.width < 0 || face.x + face.width > Sourceimg.cols ||
                face.y < 0 || face.height < 0 || face.y + face.height > Sourceimg.rows) {
                continue;
            }
            // Draw the frame in the original picture 
            rectangle(Sourceimg, face, Scalar(255, 0, 255));
#if 0
            int i = 0;
            while (true){
                // Make positive samples 
                Mat m;
                // hold img Copy the face part in to m in 
                Sourceimg(face).copyTo(m);
                // Put the face   Reset to  24x24 Size picture 
                resize(m, m, Size(24, 24));
                // Converted to grayscale 
                cvtColor(m, m, COLOR_BGR2GRAY);
                char p[200];
                sprintf(p, "F:/opencvWin/facetrain/samples/vivid/%d.jpg", i++);
                // hold mat Write as jpg file 
                imwrite(p, m);
                m.release();
                if(i == 100){
                    break;
                }
            }
#endif
        }
        imshow("camera", Sourceimg);
        //27 == ESC   wait for 30 MS exit 
        if (waitKey(30) == 27) {
            break;
        }
    }
    if (!Sourceimg.empty()) Sourceimg.release();
    if (!gray.empty()) gray.release();
    capture.release();
    tracker->stop();
    delete tracker;
    return 0;
}

Loaded above lbpcascade_frontalface.xml The face model is opencv The source code comes with , We can use this model to train our own face model , Next, I will introduce training my own face model , Open this code , When we recognize the human face, we save the face information into a size of 24*24( Pixels ) Pictures of the , keep in storage 100 Zhang , reason : I only 300 Negative sample picture （ Do not include face ）

#if 0
            int i = 0;
            while (true){
                // Make positive samples 
                Mat m;
                // hold img Copy the face part in to m in 
                Sourceimg(face).copyTo(m);
                // Put the face   Reset to  24x24 Size picture 
                resize(m, m, Size(24, 24));
                // Converted to grayscale 
                cvtColor(m, m, COLOR_BGR2GRAY);
                char p[200];
                sprintf(p, "F:/opencvWin/facetrain/samples/vivid/%d.jpg", i++);
                // hold mat Write as jpg file 
                imwrite(p, m);
                m.release();
                if(i == 100){
                    break;
                }
            }
#endif

Positive sample ：

Sort the positive sample information into xxx.xxx（ Any file name , Any file suffix ） I saved it here to vivid.data file

vivid.data The contents of the document ：

vivid/0.jpg 1 0 0 24 24
vivid/1.jpg 1 0 0 24 24
vivid/2.jpg 1 0 0 24 24
vivid/3.jpg 1 0 0 24 24
vivid/4.jpg 1 0 0 24 24
vivid/5.jpg 1 0 0 24 24
vivid/6.jpg 1 0 0 24 24
vivid/7.jpg 1 0 0 24 24
vivid/8.jpg 1 0 0 24 24
vivid/9.jpg 1 0 0 24 24

Parameter meaning ：

vivid/x.jpg Sample location （ The full path is ：F:\opencvWin\facetrain\samples\vivid\x.jpg）

1  It means there is only one face

0  0 Indicates the starting position of the face

24 24 Represents the size of the face , That is, the end position

If one picture has more than one face ,eg:2 Zhang Renren

vivid/0.jpg 2  0 0 50 50     80 80 130 130 

0 0 50 50 The starting and ending point of the first face ,80 80 130 130 The starting and ending point of the second face

Writing of document contents , Can be done in code ：

import java.io.*;
public class GeneateFile{
		public static void main(String[] args) throws Exception{//FileOutputStream 
			FileOutputStream fos = new FileOutputStream("F:/opencvWin/facetrain/samples/vivid/vivid.data");
			for(int i =0 ;i<100;i++){
				String content = String.format("vivid/%d.jpg 1 0 0 24 24\n",i);
				fos.write(content.getBytes());
			} 
			fos.close();
	}
}

take vivid.data File conversion to vec Sample file

opencv_createsamples -info vivid.data -vec vivid.vec -num 100 -w 24 -h 24
-info:  Positive sample description 
-vec :  Output positive sample file 
-num :  Number of positive samples 
-w -h:  The size of the output sample 

opencv_createsamples -info vivid.data -vec vivid.vec -num 100 -w 24 -h 24
 result ：
F:\opencvWin\facetrain\samples>opencv_createsamples -info vivid.data -vec vivid.vec -num 100 -w 24 -h 24
Info file name: vivid.data
Img file name: (NULL)
Vec file name: vivid.vec
BG  file name: (NULL)
Num: 100
BG color: 0
BG threshold: 80
Invert: FALSE
Max intensity deviation: 40
Max x angle: 1.1
Max y angle: 1.1
Max z angle: 0.5
Show samples: FALSE
Width: 24
Height: 24
Max Scale: -1
Create training samples from images collection...
Done. Created 100 samples

F:\opencvWin\facetrain\samples>

The above display indicates success .

Modify generation vivid.data Of java Code , Generating negative samples bg.data file

import java.io.*;
public class GeneateFileOpp{
		public static void main(String[] args) throws Exception{//FileOutputStream 
			FileOutputStream fos = new FileOutputStream("F:/opencvWin/facetrain/samples/bg/bg.data");
			for(int i =0 ;i<300;i++){
				String content = String.format("bg/%d.jpg\n",i);
				fos.write(content.getBytes());
			} 
			fos.close();
	}
}

bg.data The contents of the document （ Omit the part ）： Note that there is no face information

bg/0.jpg
bg/1.jpg
bg/2.jpg
bg/3.jpg
bg/4.jpg
bg/5.jpg
bg/6.jpg
bg/7.jpg
bg/8.jpg
bg/9.jpg
bg/10.jpg

Training ：

opencv_traincascade -data data -vec vivid.vec -bg bg.data -numPos 100 -numNeg 300 -numStages
15 -featureType LBP -w 24 -h 24
-data ： The target directory needs to be created manually , Used to store the generated model , The name can be customized
-vec ： Positive sample
-bg ： Negative sample
-numPos ： The number of positive samples used in the training of each classifier
-numNeg ： The number of negative samples used in the training of each classifier , Can be greater than -bg number
-numStages： Train the number of classifiers , If there are many layers , The error of classifier is smaller , But the detection speed is slow .(15-20)
-featureType: LBP
-w -h

Execution success results ：

Training until now has taken 0 days 0 hours 0 minutes 18 seconds.

===== TRAINING 7-stage =====
<BEGIN
POS count : consumed   100 : 100
NEG count : acceptanceRatio    0 : 0
Required leaf false alarm rate achieved. Branch training terminated. 

At this time in data The directory will have the following files ：

  Modify the above model loading path , You can check whether the model is valid , The operation is as follows ：