OpenCV实现人脸检测功能

时间:2021-12-05 20:04:45

本文实例为大家分享了OpenCV实现人脸检测功能的具体代码,供大家参考,具体内容如下

1、HAAR级联检测

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
#include <opencv2/opencv.hpp>
#include <iostream>
 
using namespace cv;
 
#include <iostream>
#include <cstdlib>
using namespace std;
 
int main(int artc, char** argv) {
 face_detect_haar();
 waitKey(0);
 return 0;
}
 
void face_detect_haar() {
 CascadeClassifier faceDetector;
 std::string haar_data_file = "./models/haarcascades/haarcascade_frontalface_alt_tree.xml";
 faceDetector.load(haar_data_file);
 vector<Rect> faces;
 //VideoCapture capture(0);
 VideoCapture capture("./video/test.mp4");
 Mat frame, gray;
 int count=0;
 while (capture.read(frame)) {
 int64 start = getTickCount();
 if (frame.empty())
 {
 break;
 }
 // 水平镜像调整
 // flip(frame, frame, 1);
 imshow("input", frame);
 if (frame.channels() == 4)
 cvtColor(frame, frame, COLOR_BGRA2BGR);
 cvtColor(frame, gray, COLOR_BGR2GRAY);
 equalizeHist(gray, gray);
 faceDetector.detectMultiScale(gray, faces, 1.2, 1, 0, Size(30, 30), Size(400, 400));
 for (size_t t = 0; t < faces.size(); t++) {
 count++;
 rectangle(frame, faces[t], Scalar(0, 255, 0), 2, 8, 0);
 }
 float fps = getTickFrequency() / (getTickCount() - start);
 ostringstream ss;ss.str("");
 ss << "FPS: " << fps << " ; inference time: " << time << " ms";
 putText(frame, ss.str(), Point(20, 20), 0, 0.75, Scalar(0, 0, 255), 2, 8);
 imshow("haar_face_detection", frame);
 if (waitKey(1) >= 0) break;
 }
 
 printf("total face: %d\n", count);
}

2、 DNN人脸检测

?
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
#include <opencv2/dnn.hpp>
#include <opencv2/opencv.hpp>
 
using namespace cv;
using namespace cv::dnn;
 
#include <iostream>
#include <cstdlib>
using namespace std;
 
const size_t inWidth = 300;
const size_t inHeight = 300;
const double inScaleFactor = 1.0;
const Scalar meanVal(104.0, 177.0, 123.0);
const float confidenceThreshold = 0.7;
void face_detect_dnn();
void mtcnn_demo();
int main(int argc, char** argv)
{
 face_detect_dnn();
 waitKey(0);
 return 0;
}
 
void face_detect_dnn() {
 //这里采用tensorflow模型
 std::string modelBinary = "./models/dnn/face_detector/opencv_face_detector_uint8.pb";
 std::string modelDesc = "./models/dnn/face_detector/opencv_face_detector.pbtxt";
 // 初始化网络
 dnn::Net net = readNetFromTensorflow(modelBinary, modelDesc);
 
 net.setPreferableBackend(DNN_BACKEND_OPENCV);
 net.setPreferableTarget(DNN_TARGET_CPU);
 if (net.empty())
 {
  printf("Load models fail...\n");
  return;
 }
 
 // 打开摄像头
 // VideoCapture capture(0);
 VideoCapture capture("./video/test.mp4");
 if (!capture.isOpened()) {
  printf("Don't find video...\n");
  return;
 }
 
 Mat frame;
 int count=0;
 while (capture.read(frame)) {
  int64 start = getTickCount();
  if (frame.empty())
  {
   break;
  }
  // 水平镜像调整
  // flip(frame, frame, 1);
  imshow("input", frame);
  if (frame.channels() == 4)
   cvtColor(frame, frame, COLOR_BGRA2BGR);
 
  // 输入数据调整
  Mat inputBlob = blobFromImage(frame, inScaleFactor,
   Size(inWidth, inHeight), meanVal, false, false);
  net.setInput(inputBlob, "data");
 
  // 人脸检测
  Mat detection = net.forward("detection_out");
  vector<double> layersTimings;
  double freq = getTickFrequency() / 1000;
  double time = net.getPerfProfile(layersTimings) / freq;
  Mat detectionMat(detection.size[2], detection.size[3], CV_32F, detection.ptr<float>());
 
  ostringstream ss;
  for (int i = 0; i < detectionMat.rows; i++)
  {
   // 置信度 0~1之间
   float confidence = detectionMat.at<float>(i, 2);
   if (confidence > confidenceThreshold)
   {
    count++;
    int xLeftBottom = static_cast<int>(detectionMat.at<float>(i, 3) * frame.cols);
    int yLeftBottom = static_cast<int>(detectionMat.at<float>(i, 4) * frame.rows);
    int xRightTop = static_cast<int>(detectionMat.at<float>(i, 5) * frame.cols);
    int yRightTop = static_cast<int>(detectionMat.at<float>(i, 6) * frame.rows);
 
    Rect object((int)xLeftBottom, (int)yLeftBottom,
     (int)(xRightTop - xLeftBottom),
     (int)(yRightTop - yLeftBottom));
 
    rectangle(frame, object, Scalar(0, 255, 0));
 
    ss << confidence;
    std::string conf(ss.str());
    std::string label = "Face: " + conf;
    int baseLine = 0;
    Size labelSize = getTextSize(label, FONT_HERSHEY_SIMPLEX, 0.5, 1, &baseLine);
    rectangle(frame, Rect(Point(xLeftBottom, yLeftBottom - labelSize.height),
     Size(labelSize.width, labelSize.height + baseLine)),
     Scalar(255, 255, 255), FILLED);
    putText(frame, label, Point(xLeftBottom, yLeftBottom),
     FONT_HERSHEY_SIMPLEX, 0.5, Scalar(0, 0, 0));
   }
  }
  float fps = getTickFrequency() / (getTickCount() - start);
  ss.str("");
  ss << "FPS: " << fps << " ; inference time: " << time << " ms";
  putText(frame, ss.str(), Point(20, 20), 0, 0.75, Scalar(0, 0, 255), 2, 8);
  imshow("dnn_face_detection", frame);
  if (waitKey(1) >= 0) break;
 }
 printf("total face: %d\n", count);
}

以上就是本文的全部内容,希望对大家的学习有所帮助,也希望大家多多支持服务器之家。

原文链接:https://blog.csdn.net/u012156872/article/details/104298472