C# OpenCvSharp DNN FreeYOLO 人脸检测&人脸图像质量评估

目录

效果

模型信息

yolo_free_huge_widerface_192x320.onnx

face-quality-assessment.onnx

项目

代码

frmMain.cs

FreeYoloFace

FaceQualityAssessment.cs

下载

C# OpenCvSharp DNN FreeYOLO 人脸检测&人脸图像质量评估

效果

模型信息

yolo_free_huge_widerface_192x320.onnx


Inputs
-------------------------
name：input
tensor：Float[1, 3, 192, 320]
---------------------------------------------------------------

Outputs
-------------------------
name：output
tensor：Float[1, 1260, 6]
---------------------------------------------------------------

face-quality-assessment.onnx

Inputs
-------------------------
name：input
tensor：Float[1, 3, 112, 112]
---------------------------------------------------------------

Outputs
-------------------------
name：quality
tensor：Float[1, 10]
---------------------------------------------------------------

项目

代码

frmMain.cs

using OpenCvSharp;
using System;
using System.Collections.Generic;
using System.Drawing;
using System.Text;
using System.Windows.Forms;

namespace OpenCvSharp_DNN_Demo
{
    public partial class frmMain : Form
    {
        public frmMain()
        {
            InitializeComponent();
        }

        string fileFilter = "*.*|*.bmp;*.jpg;*.jpeg;*.tiff;*.tiff;*.png";
        string image_path = "";

        DateTime dt1 = DateTime.Now;
        DateTime dt2 = DateTime.Now;

        StringBuilder sb = new StringBuilder();

        Mat image;
        Mat result_image;

        FaceQualityAssessment fqa = new FaceQualityAssessment("model/face-quality-assessment.onnx");
        FreeYoloFace face = new FreeYoloFace("model/yolo_free_huge_widerface_192x320.onnx");

        private void button1_Click(object sender, EventArgs e)
        {
            OpenFileDialog ofd = new OpenFileDialog();
            ofd.Filter = fileFilter;
            if (ofd.ShowDialog() != DialogResult.OK) return;

            pictureBox1.Image = null;
            pictureBox2.Image = null;
            textBox1.Text = "";

            image_path = ofd.FileName;
            pictureBox1.Image = new Bitmap(image_path);
            image = new Mat(image_path);
        }

        private void Form1_Load(object sender, EventArgs e)
        {
            image_path = "test_img/1.jpg";
            pictureBox1.Image = new Bitmap(image_path);
        }

        private unsafe void button2_Click(object sender, EventArgs e)
        {
            if (image_path == "")
            {
                return;
            }
            textBox1.Text = "检测中，请稍等……";
            if (pictureBox2.Image != null)
            {
                pictureBox2.Image.Dispose();
            }
            pictureBox2.Image = null;
            sb.Clear();
            Application.DoEvents();

            image = new Mat(image_path);

            dt1 = DateTime.Now;
            List ltFace = face.Detect(image);
            dt2 = DateTime.Now;

            if (ltFace.Count > 0)
            {
                sb.AppendLine("推理耗时:" + (dt2 - dt1).TotalMilliseconds + "ms");
                result_image = image.Clone();
                foreach (var item in ltFace)
                {
                    Mat crop_img = new Mat(image, item.rect);
                    float fqa_prob_mean = fqa.Detect(crop_img);
                    crop_img.Dispose();
                    Cv2.Rectangle(result_image, new OpenCvSharp.Point(item.rect.X, item.rect.Y), new OpenCvSharp.Point(item.rect.X + item.rect.Width, item.rect.Y + item.rect.Height), new Scalar(0, 0, 255), 2);
                    string label = "prob:" + item.prob.ToString("0.00") + " fqa_score:" + fqa_prob_mean.ToString("0.00");
                    sb.AppendLine(label);
                    Cv2.PutText(result_image, label, new OpenCvSharp.Point(item.rect.X, item.rect.Y - 5), HersheyFonts.HersheySimplex, 1, new Scalar(0, 0, 255), 2);
                }
                pictureBox2.Image = new Bitmap(result_image.ToMemoryStream());
                textBox1.Text = sb.ToString();
            }
            else
            {
                textBox1.Text = "未检测到人脸";
            }
        }

        private void pictureBox2_DoubleClick(object sender, EventArgs e)
        {
            Common.ShowNormalImg(pictureBox2.Image);
        }

        private void pictureBox1_DoubleClick(object sender, EventArgs e)
        {
            Common.ShowNormalImg(pictureBox1.Image);
        }
    }
}

FreeYoloFace.cs

  1. using OpenCvSharp.Dnn;
  2. using OpenCvSharp;
  3. using System;
  4. using System.Collections.Generic;
  5. using System.Linq;
  6.  
  7. namespace OpenCvSharp_DNN_Demo
  8. {
  9.     public class FreeYoloFace
  10.     {
  11.  
  12.         float confThreshold;
  13.         float nmsThreshold;
  14.  
  15.         int num_stride = 3;
  16.         float[] strides = new float[3] { 8.0f, 16.0f, 32.0f };
  17.  
  18.         string modelpath;
  19.  
  20.         int inpHeight;
  21.         int inpWidth;
  22.  
  23.         List<string> class_names;
  24.         int num_class;
  25.  
  26.         Net opencv_net;
  27.         Mat BN_image;
  28.  
  29.         Mat image;
  30.  
  31.         public FreeYoloFace(string modelpath)
  32.         {
  33.             opencv_net = CvDnn.ReadNetFromOnnx(modelpath);
  34.  
  35.             class_names = new List<string> { "face" };
  36.             num_class = 1;
  37.  
  38.             confThreshold = 0.8f;
  39.             nmsThreshold = 0.5f;
  40.  
  41.             inpHeight = 192;
  42.             inpWidth = 320;
  43.         }
  44.  
  45.  
  46.         unsafe public List<Face> Detect(Mat image)
  47.         {
  48.             List<Face> ltFace = new List<Face>();
  49.             float ratio = Math.Min(1.0f * inpHeight / image.Rows, 1.0f * inpWidth / image.Cols);
  50.             int neww = (int)(image.Cols * ratio);
  51.             int newh = (int)(image.Rows * ratio);
  52.  
  53.             Mat dstimg = new Mat();
  54.             Cv2.Resize(image, dstimg, new OpenCvSharp.Size(neww, newh));
  55.             Cv2.CopyMakeBorder(dstimg, dstimg, 0, inpHeight - newh, 0, inpWidth - neww, BorderTypes.Constant);
  56.             BN_image = CvDnn.BlobFromImage(dstimg);
  57.  
  58.             //配置图片输入数据
  59.             opencv_net.SetInput(BN_image);
  60.  
  61.             //模型推理，读取推理结果
  62.             Mat[] outs = new Mat[1] { new Mat() };
  63.             string[] outBlobNames = opencv_net.GetUnconnectedOutLayersNames().ToArray();
  64.  
  65.             opencv_net.Forward(outs, outBlobNames);
  66.  
  67.             int num_proposal = outs[0].Size(1);
  68.             int nout = outs[0].Size(2);
  69.  
  70.             float* pdata = (float*)outs[0].Data;
  71.  
  72.             List<float> confidences = new List<float>();
  73.             List<Rect> boxes = new List<Rect>();
  74.             List<int> classIds = new List<int>();
  75.  
  76.             for (int n = 0; n < num_stride; n++)
  77.             {
  78.                 int num_grid_x = (int)Math.Ceiling(inpWidth / strides[n]);
  79.                 int num_grid_y = (int)Math.Ceiling(inpHeight / strides[n]);
  80.  
  81.                 for (int i = 0; i < num_grid_y; i++)
  82.                 {
  83.                     for (int j = 0; j < num_grid_x; j++)
  84.                     {
  85.                         float box_score = pdata[4];
  86.                         int max_ind = 0;
  87.                         float max_class_socre = 0;
  88.                         for (int k = 0; k < num_class; k++)
  89.                         {
  90.                             if (pdata[k + 5] > max_class_socre)
  91.                             {
  92.                                 max_class_socre = pdata[k + 5];
  93.                                 max_ind = k;
  94.                             }
  95.                         }
  96.                         max_class_socre = max_class_socre * box_score;
  97.                         max_class_socre = (float)Math.Sqrt(max_class_socre);
  98.  
  99.                         if (max_class_socre > confThreshold)
  100.                         {
  101.                             float cx = (0.5f + j + pdata[0]) * strides[n];  //cx
  102.                             float cy = (0.5f + i + pdata[1]) * strides[n];   //cy
  103.                             float w = (float)(Math.Exp(pdata[2]) * strides[n]);   //w
  104.                             float h = (float)(Math.Exp(pdata[3]) * strides[n]);  //h
  105.  
  106.                             float xmin = (float)((cx - 0.5 * w) / ratio);
  107.                             float ymin = (float)((cy - 0.5 * h) / ratio);
  108.                             float xmax = (float)((cx + 0.5 * w) / ratio);
  109.                             float ymax = (float)((cy + 0.5 * h) / ratio);
  110.  
  111.                             int left = (int)((cx - 0.5 * w) / ratio);
  112.                             int top = (int)((cy - 0.5 * h) / ratio);
  113.                             int width = (int)(w / ratio);
  114.                             int height = (int)(h / ratio);
  115.  
  116.                             confidences.Add(max_class_socre);
  117.                             boxes.Add(new Rect(left, top, width, height));
  118.                             classIds.Add(max_ind);
  119.                         }
  120.                         pdata += nout;
  121.                     }
  122.                 }
  123.  
  124.             }
  125.  
  126.             int[] indices;
  127.             CvDnn.NMSBoxes(boxes, confidences, confThreshold, nmsThreshold, out indices);
  128.  
  129.             for (int ii = 0; ii < indices.Length; ++ii)
  130.             {
  131.                 int idx = indices[ii];
  132.                 Rect box = boxes[idx];
  133.                 ltFace.Add(new Face(box, confidences[idx]));
  134.             }
  135.  
  136.             outs[0].Dispose();
  137.             BN_image.Dispose();
  138.             dstimg.Dispose();
  139.  
  140.             return ltFace;
  141.         }
  142.     }
  143. }

FaceQualityAssessment.cs

  1. using OpenCvSharp;
  2. using OpenCvSharp.Dnn;
  3. using System.Linq;
  4.  
  5. namespace OpenCvSharp_DNN_Demo
  6. {
  7.     public class FaceQualityAssessment
  8.     {
  9.         Net net;
  10.  
  11.         int inpWidth = 112;
  12.         int inpHeight = 112;
  13.  
  14.         float[] mean = new float[] { 0.5f, 0.5f, 0.5f };
  15.  
  16.         float[] std = new float[] { 0.5f, 0.5f, 0.5f };
  17.  
  18.         public FaceQualityAssessment(string modelpath)
  19.         {
  20.             net = CvDnn.ReadNetFromOnnx(modelpath);
  21.         }
  22.  
  23.         unsafe public float Detect(Mat cropped)
  24.         {
  25.             Mat rgbimg = new Mat();
  26.             Cv2.CvtColor(cropped, rgbimg, ColorConversionCodes.BGR2RGB);
  27.             Cv2.Resize(rgbimg, rgbimg, new Size(inpWidth, inpHeight));
  28.             Mat normalized_mat = Normalize(rgbimg);
  29.  
  30.             Mat blob = CvDnn.BlobFromImage(normalized_mat);
  31.  
  32.             //配置图片输入数据
  33.             net.SetInput(blob);
  34.  
  35.             //模型推理，读取推理结果
  36.             Mat[] outs = new Mat[1] { new Mat() };
  37.             string[] outBlobNames = net.GetUnconnectedOutLayersNames().ToArray();
  38.  
  39.             net.Forward(outs, outBlobNames);
  40.  
  41.             float* pdata = (float*)outs[0].Data;  //形状1x10
  42.             int length = outs[0].Size(1);
  43.             float fqa_prob_mean = 0;
  44.             for (int i = 0; i < length; i++)
  45.             {
  46.                 fqa_prob_mean += pdata[i];
  47.             }
  48.             fqa_prob_mean /= length;
  49.  
  50.             rgbimg.Dispose();
  51.             normalized_mat.Dispose();
  52.             blob.Dispose();
  53.             outs[0].Dispose();
  54.             return fqa_prob_mean;
  55.         }
  56.  
  57.         Mat Normalize(Mat src)
  58.         {
  59.             Mat[] bgr = src.Split();
  60.             for (int i = 0; i < bgr.Length; ++i)
  61.             {
  62.                 bgr[i].ConvertTo(bgr[i], MatType.CV_32FC1, 1.0 / (255.0 * std[i]), (0.0 - mean[i]) / std[i]);
  63.             }
  64.             Cv2.Merge(bgr, src);
  65.             foreach (Mat channel in bgr)
  66.             {
  67.                 channel.Dispose();
  68.             }
  69.             return src;
  70.         }
  71.  
  72.     }
  73. }

下载

源码下载

天天代码码天天
微信公众号
.NET 人工智能实践
注：本文转载自blog.csdn.net的天天代码码天天的文章"https://lw112190.blog.csdn.net/article/details/135472428"。版权归原作者所有，此博客不拥有其著作权，亦不承担相应法律责任。如有侵权，请联系我们删除。
复制链接
相关推荐
发表评论

评论记录：

未查询到任何数据！