C# yolov8 TensorRT Demo

C# yolov8 TensorRT Demo

目录

效果

说明 

项目

代码

下载

效果

说明 

环境

NVIDIA GeForce RTX 4060 Laptop GPU

cuda12.1+cudnn 8.8.1+TensorRT-8.6.1.6

版本和我不一致的需要重新编译TensorRtExtern.dll，TensorRtExtern源码地址：https://github.com/guojin-yan/TensorRT-CSharp-API/tree/TensorRtSharp2.0/src/TensorRtExtern

Windows版 CUDA安装参考：http://iyenn.com/rec/1661921.html

项目

代码

Form2.cs

using OpenCvSharp;
using System;
using System.Collections.Generic;
using System.Diagnostics;
using System.Drawing;
using System.IO;
using System.Threading;
using System.Windows.Forms;
using TensorRtSharp.Custom;

namespace yolov8_TensorRT_Demo
{
    public partial class Form2 : Form
    {
        public Form2()
        {
            InitializeComponent();
        }

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

        YoloV8 yoloV8;
        Mat image;

        string image_path = "";
        string model_path;

        string video_path = "";
        string videoFilter = "*.mp4|*.mp4;";
        VideoCapture vcapture;
        VideoWriter vwriter;
        bool saveDetVideo = false;


        ///


        /// 单图推理
        ///
        ///
        ///
        private void button2_Click(object sender, EventArgs e)
        {

            if (image_path == "")
            {
                return;
            }

            button2.Enabled = false;
            pictureBox2.Image = null;
            textBox1.Text = "";

            Application.DoEvents();

            image = new Mat(image_path);

            List detResults = yoloV8.Detect(image);

            //绘制结果
            Mat result_image = image.Clone();
            foreach (DetectionResult r in detResults)
            {
                Cv2.PutText(result_image, $"{r.Class}:{r.Confidence:P0}", new OpenCvSharp.Point(r.Rect.TopLeft.X, r.Rect.TopLeft.Y - 10), HersheyFonts.HersheySimplex, 1, Scalar.Red, 2);
                Cv2.Rectangle(result_image, r.Rect, Scalar.Red, thickness: 2);
            }

            if (pictureBox2.Image != null)
            {
                pictureBox2.Image.Dispose();
            }
            pictureBox2.Image = new Bitmap(result_image.ToMemoryStream());
            textBox1.Text = yoloV8.DetectTime();

            button2.Enabled = true;

        }

        ///


        /// 窗体加载，初始化
        ///
        ///
        ///
        private void Form1_Load(object sender, EventArgs e)
        {
            image_path = "test/zidane.jpg";
            pictureBox1.Image = new Bitmap(image_path);

            model_path = "model/yolov8n.engine";

            if (!File.Exists(model_path))
            {
                //有点耗时，需等待
                Nvinfer.OnnxToEngine("model/yolov8n.onnx", 20);
            }

            yoloV8 = new YoloV8(model_path, "model/lable.txt");

        }

        ///


        /// 选择图片
        ///
        ///
        ///
        private void button1_Click_1(object sender, EventArgs e)
        {
            OpenFileDialog ofd = new OpenFileDialog();
            ofd.Filter = imgFilter;
            if (ofd.ShowDialog() != DialogResult.OK) return;

            pictureBox1.Image = null;

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

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

        ///


        /// 选择视频
        ///
        ///
        ///
        private void button4_Click(object sender, EventArgs e)
        {
            OpenFileDialog ofd = new OpenFileDialog();
            ofd.Filter = videoFilter;
            ofd.InitialDirectory = Application.StartupPath + "\\test";

            if (ofd.ShowDialog() != DialogResult.OK) return;

            video_path = ofd.FileName;

            button3_Click(null, null);

        }

        ///


        /// 视频推理
        ///
        ///
        ///
        private void button3_Click(object sender, EventArgs e)
        {
            if (video_path == null)
            {
                return;
            }

            textBox1.Text = "开始检测";

            Application.DoEvents();

            Thread thread = new Thread(new ThreadStart(VideoDetection));

            thread.Start();
            thread.Join();

            textBox1.Text = "检测完成！";
        }

        void VideoDetection()
        {
            vcapture = new VideoCapture(video_path);
            if (!vcapture.IsOpened())
            {
                MessageBox.Show("打开视频文件失败");
                return;
            }

            Mat frame = new Mat();
            List detResults;

            // 获取视频的fps
            double videoFps = vcapture.Get(VideoCaptureProperties.Fps);
            // 计算等待时间（毫秒）
            int delay = (int)(1000 / videoFps);
            Stopwatch _stopwatch = new Stopwatch();

            if (checkBox1.Checked)
            {
                vwriter = new VideoWriter("out.mp4", FourCC.X264, vcapture.Fps, new OpenCvSharp.Size(vcapture.FrameWidth, vcapture.FrameHeight));
                saveDetVideo = true;
            }
            else {
                saveDetVideo = false;
            }

            while (vcapture.Read(frame))
            {
                if (frame.Empty())
                {
                    MessageBox.Show("读取失败");
                    return;
                }

                _stopwatch.Restart();

                delay = (int)(1000 / videoFps);

                detResults = yoloV8.Detect(frame);

                //绘制结果
                foreach (DetectionResult r in detResults)
                {
                    Cv2.PutText(frame, $"{r.Class}:{r.Confidence:P0}", new OpenCvSharp.Point(r.Rect.TopLeft.X, r.Rect.TopLeft.Y - 10), HersheyFonts.HersheySimplex, 1, Scalar.Red, 2);
                    Cv2.Rectangle(frame, r.Rect, Scalar.Red, thickness: 2);
                }
                Cv2.PutText(frame, "preprocessTime:" + yoloV8.preprocessTime.ToString("F2")+"ms", new OpenCvSharp.Point(10, 30), HersheyFonts.HersheySimplex, 1, Scalar.Red, 2);
                Cv2.PutText(frame, "inferTime:" + yoloV8.inferTime.ToString("F2") + "ms", new OpenCvSharp.Point(10, 70), HersheyFonts.HersheySimplex, 1, Scalar.Red, 2);
                Cv2.PutText(frame, "postprocessTime:" + yoloV8.postprocessTime.ToString("F2") + "ms", new OpenCvSharp.Point(10, 110), HersheyFonts.HersheySimplex, 1, Scalar.Red, 2);
                Cv2.PutText(frame, "totalTime:" + yoloV8.totalTime.ToString("F2") + "ms", new OpenCvSharp.Point(10, 150), HersheyFonts.HersheySimplex, 1, Scalar.Red, 2);
                Cv2.PutText(frame, "video fps:" + videoFps.ToString("F2"), new OpenCvSharp.Point(10, 190), HersheyFonts.HersheySimplex, 1, Scalar.Red, 2);
                Cv2.PutText(frame, "det fps:" + yoloV8.detFps.ToString("F2"), new OpenCvSharp.Point(10, 230), HersheyFonts.HersheySimplex, 1, Scalar.Red, 2);

                if (saveDetVideo)
                {
                    vwriter.Write(frame);
                }

                Cv2.ImShow("DetectionResult", frame);

                // for test
                // delay = 1;

                delay = (int)(delay - _stopwatch.ElapsedMilliseconds);
                if (delay <= 0)
                {
                    delay = 1;
                }
                //Console.WriteLine("delay:" + delay.ToString()) ;
                if (Cv2.WaitKey(delay) == 27)
                {
                    break; // 如果按下ESC，退出循环
                }
            }

            Cv2.DestroyAllWindows();
            vcapture.Release();
            if (saveDetVideo)
            {
                vwriter.Release();
            }

        }
    }

}
 

  1. using OpenCvSharp;
  2. using System;
  3. using System.Collections.Generic;
  4. using System.Diagnostics;
  5. using System.Drawing;
  6. using System.IO;
  7. using System.Threading;
  8. using System.Windows.Forms;
  9. using TensorRtSharp.Custom;
  10.  
  11. namespace yolov8_TensorRT_Demo
  12. {
  13.     public partial class Form2 : Form
  14.     {
  15.         public Form2()
  16.         {
  17.             InitializeComponent();
  18.         }
  19.  
  20.         string imgFilter = "*.*|*.bmp;*.jpg;*.jpeg;*.tiff;*.tiff;*.png";
  21.  
  22.         YoloV8 yoloV8;
  23.         Mat image;
  24.  
  25.         string image_path = "";
  26.         string model_path;
  27.  
  28.         string video_path = "";
  29.         string videoFilter = "*.mp4|*.mp4;";
  30.         VideoCapture vcapture;
  31.         VideoWriter vwriter;
  32.         bool saveDetVideo = false;
  33.  
  34.  
  35.         /// <summary>
  36.         /// 单图推理
  37.         /// </summary>
  38.         /// <param name="sender"></param>
  39.         /// <param name="e"></param>
  40.         private void button2_Click(object sender, EventArgs e)
  41.         {
  42.  
  43.             if (image_path == "")
  44.             {
  45.                 return;
  46.             }
  47.  
  48.             button2.Enabled = false;
  49.             pictureBox2.Image = null;
  50.             textBox1.Text = "";
  51.  
  52.             Application.DoEvents();
  53.  
  54.             image = new Mat(image_path);
  55.  
  56.             List<DetectionResult> detResults = yoloV8.Detect(image);
  57.  
  58.             //绘制结果
  59.             Mat result_image = image.Clone();
  60.             foreach (DetectionResult r in detResults)
  61.             {
  62.                 Cv2.PutText(result_image, $"{r.Class}:{r.Confidence:P0}", new OpenCvSharp.Point(r.Rect.TopLeft.X, r.Rect.TopLeft.Y - 10), HersheyFonts.HersheySimplex, 1, Scalar.Red, 2);
  63.                 Cv2.Rectangle(result_image, r.Rect, Scalar.Red, thickness: 2);
  64.             }
  65.  
  66.             if (pictureBox2.Image != null)
  67.             {
  68.                 pictureBox2.Image.Dispose();
  69.             }
  70.             pictureBox2.Image = new Bitmap(result_image.ToMemoryStream());
  71.             textBox1.Text = yoloV8.DetectTime();
  72.  
  73.             button2.Enabled = true;
  74.  
  75.         }
  76.  
  77.         /// <summary>
  78.         /// 窗体加载，初始化
  79.         /// </summary>
  80.         /// <param name="sender"></param>
  81.         /// <param name="e"></param>
  82.         private void Form1_Load(object sender, EventArgs e)
  83.         {
  84.             image_path = "test/zidane.jpg";
  85.             pictureBox1.Image = new Bitmap(image_path);
  86.  
  87.             model_path = "model/yolov8n.engine";
  88.  
  89.             if (!File.Exists(model_path))
  90.             {
  91.                 //有点耗时，需等待
  92.                 Nvinfer.OnnxToEngine("model/yolov8n.onnx", 20);
  93.             }
  94.  
  95.             yoloV8 = new YoloV8(model_path, "model/lable.txt");
  96.  
  97.         }
  98.  
  99.         /// <summary>
  100.         /// 选择图片
  101.         /// </summary>
  102.         /// <param name="sender"></param>
  103.         /// <param name="e"></param>
  104.         private void button1_Click_1(object sender, EventArgs e)
  105.         {
  106.             OpenFileDialog ofd = new OpenFileDialog();
  107.             ofd.Filter = imgFilter;
  108.             if (ofd.ShowDialog() != DialogResult.OK) return;
  109.  
  110.             pictureBox1.Image = null;
  111.  
  112.             image_path = ofd.FileName;
  113.             pictureBox1.Image = new Bitmap(image_path);
  114.  
  115.             textBox1.Text = "";
  116.             pictureBox2.Image = null;
  117.         }
  118.  
  119.         /// <summary>
  120.         /// 选择视频
  121.         /// </summary>
  122.         /// <param name="sender"></param>
  123.         /// <param name="e"></param>
  124.         private void button4_Click(object sender, EventArgs e)
  125.         {
  126.             OpenFileDialog ofd = new OpenFileDialog();
  127.             ofd.Filter = videoFilter;
  128.             ofd.InitialDirectory = Application.StartupPath + "\\test";
  129.  
  130.             if (ofd.ShowDialog() != DialogResult.OK) return;
  131.  
  132.             video_path = ofd.FileName;
  133.  
  134.             button3_Click(null, null);
  135.  
  136.         }
  137.  
  138.         /// <summary>
  139.         /// 视频推理
  140.         /// </summary>
  141.         /// <param name="sender"></param>
  142.         /// <param name="e"></param>
  143.         private void button3_Click(object sender, EventArgs e)
  144.         {
  145.             if (video_path == null)
  146.             {
  147.                 return;
  148.             }
  149.  
  150.             textBox1.Text = "开始检测";
  151.  
  152.             Application.DoEvents();
  153.  
  154.             Thread thread = new Thread(new ThreadStart(VideoDetection));
  155.  
  156.             thread.Start();
  157.             thread.Join();
  158.  
  159.             textBox1.Text = "检测完成！";
  160.         }
  161.  
  162.         void VideoDetection()
  163.         {
  164.             vcapture = new VideoCapture(video_path);
  165.             if (!vcapture.IsOpened())
  166.             {
  167.                 MessageBox.Show("打开视频文件失败");
  168.                 return;
  169.             }
  170.  
  171.             Mat frame = new Mat();
  172.             List<DetectionResult> detResults;
  173.  
  174.             // 获取视频的fps
  175.             double videoFps = vcapture.Get(VideoCaptureProperties.Fps);
  176.             // 计算等待时间（毫秒）
  177.             int delay = (int)(1000 / videoFps);
  178.             Stopwatch _stopwatch = new Stopwatch();
  179.  
  180.             if (checkBox1.Checked)
  181.             {
  182.                 vwriter = new VideoWriter("out.mp4", FourCC.X264, vcapture.Fps, new OpenCvSharp.Size(vcapture.FrameWidth, vcapture.FrameHeight));
  183.                 saveDetVideo = true;
  184.             }
  185.             else {
  186.                 saveDetVideo = false;
  187.             }
  188.  
  189.             while (vcapture.Read(frame))
  190.             {
  191.                 if (frame.Empty())
  192.                 {
  193.                     MessageBox.Show("读取失败");
  194.                     return;
  195.                 }
  196.  
  197.                 _stopwatch.Restart();
  198.  
  199.                 delay = (int)(1000 / videoFps);
  200.  
  201.                 detResults = yoloV8.Detect(frame);
  202.  
  203.                 //绘制结果
  204.                 foreach (DetectionResult r in detResults)
  205.                 {
  206.                     Cv2.PutText(frame, $"{r.Class}:{r.Confidence:P0}", new OpenCvSharp.Point(r.Rect.TopLeft.X, r.Rect.TopLeft.Y - 10), HersheyFonts.HersheySimplex, 1, Scalar.Red, 2);
  207.                     Cv2.Rectangle(frame, r.Rect, Scalar.Red, thickness: 2);
  208.                 }
  209.                 Cv2.PutText(frame, "preprocessTime:" + yoloV8.preprocessTime.ToString("F2")+"ms", new OpenCvSharp.Point(10, 30), HersheyFonts.HersheySimplex, 1, Scalar.Red, 2);
  210.                 Cv2.PutText(frame, "inferTime:" + yoloV8.inferTime.ToString("F2") + "ms", new OpenCvSharp.Point(10, 70), HersheyFonts.HersheySimplex, 1, Scalar.Red, 2);
  211.                 Cv2.PutText(frame, "postprocessTime:" + yoloV8.postprocessTime.ToString("F2") + "ms", new OpenCvSharp.Point(10, 110), HersheyFonts.HersheySimplex, 1, Scalar.Red, 2);
  212.                 Cv2.PutText(frame, "totalTime:" + yoloV8.totalTime.ToString("F2") + "ms", new OpenCvSharp.Point(10, 150), HersheyFonts.HersheySimplex, 1, Scalar.Red, 2);
  213.                 Cv2.PutText(frame, "video fps:" + videoFps.ToString("F2"), new OpenCvSharp.Point(10, 190), HersheyFonts.HersheySimplex, 1, Scalar.Red, 2);
  214.                 Cv2.PutText(frame, "det fps:" + yoloV8.detFps.ToString("F2"), new OpenCvSharp.Point(10, 230), HersheyFonts.HersheySimplex, 1, Scalar.Red, 2);
  215.  
  216.                 if (saveDetVideo)
  217.                 {
  218.                     vwriter.Write(frame);
  219.                 }
  220.  
  221.                 Cv2.ImShow("DetectionResult", frame);
  222.  
  223.                 // for test
  224.                 // delay = 1;
  225.  
  226.                 delay = (int)(delay - _stopwatch.ElapsedMilliseconds);
  227.                 if (delay <= 0)
  228.                 {
  229.                     delay = 1;
  230.                 }
  231.                 //Console.WriteLine("delay:" + delay.ToString()) ;
  232.                 if (Cv2.WaitKey(delay) == 27)
  233.                 {
  234.                     break; // 如果按下ESC，退出循环
  235.                 }
  236.             }
  237.  
  238.             Cv2.DestroyAllWindows();
  239.             vcapture.Release();
  240.             if (saveDetVideo)
  241.             {
  242.                 vwriter.Release();
  243.             }
  244.  
  245.         }
  246.     }
  247.  
  248. }

YoloV8.cs

  1. using OpenCvSharp;
  2. using OpenCvSharp.Dnn;
  3. using System;
  4. using System.Collections.Generic;
  5. using System.Diagnostics;
  6. using System.IO;
  7. using System.Linq;
  8. using System.Text;
  9. using TensorRtSharp.Custom;
  10.  
  11. namespace yolov8_TensorRT_Demo
  12. {
  13.     public class YoloV8
  14.     {
  15.  
  16.         float[] input_tensor_data;
  17.         float[] outputData;
  18.         List<DetectionResult> detectionResults;
  19.  
  20.         int input_height;
  21.         int input_width;
  22.  
  23.         Nvinfer predictor;
  24.  
  25.         string[] class_names;
  26.         int class_num;
  27.         int box_num;
  28.  
  29.         float conf_threshold;
  30.         float nms_threshold;
  31.  
  32.         float ratio_height;
  33.         float ratio_width;
  34.  
  35.         public double preprocessTime;
  36.         public double inferTime;
  37.         public double postprocessTime;
  38.         public double totalTime;
  39.         public double detFps;
  40.  
  41.         public String DetectTime()
  42.         {
  43.             StringBuilder stringBuilder = new StringBuilder();
  44.             stringBuilder.AppendLine($"Preprocess: {preprocessTime:F2}ms");
  45.             stringBuilder.AppendLine($"Infer: {inferTime:F2}ms");
  46.             stringBuilder.AppendLine($"Postprocess: {postprocessTime:F2}ms");
  47.             stringBuilder.AppendLine($"Total: {totalTime:F2}ms");
  48.  
  49.             return stringBuilder.ToString();
  50.         }
  51.  
  52.         public YoloV8(string model_path, string classer_path)
  53.         {
  54.             predictor = new Nvinfer(model_path);
  55.  
  56.             class_names = File.ReadAllLines(classer_path, Encoding.UTF8);
  57.             class_num = class_names.Length;
  58.  
  59.             input_height = 640;
  60.             input_width = 640;
  61.  
  62.             box_num = 8400;
  63.  
  64.             conf_threshold = 0.25f;
  65.             nms_threshold = 0.5f;
  66.  
  67.             detectionResults = new List<DetectionResult>();
  68.         }
  69.  
  70.         void Preprocess(Mat image)
  71.         {
  72.             //图片缩放
  73.             int height = image.Rows;
  74.             int width = image.Cols;
  75.             Mat temp_image = image.Clone();
  76.             if (height > input_height || width > input_width)
  77.             {
  78.                 float scale = Math.Min((float)input_height / height, (float)input_width / width);
  79.                 OpenCvSharp.Size new_size = new OpenCvSharp.Size((int)(width * scale), (int)(height * scale));
  80.                 Cv2.Resize(image, temp_image, new_size);
  81.             }
  82.             ratio_height = (float)height / temp_image.Rows;
  83.             ratio_width = (float)width / temp_image.Cols;
  84.             Mat input_img = new Mat();
  85.             Cv2.CopyMakeBorder(temp_image, input_img, 0, input_height - temp_image.Rows, 0, input_width - temp_image.Cols, BorderTypes.Constant, 0);
  86.  
  87.             //归一化
  88.             input_img.ConvertTo(input_img, MatType.CV_32FC3, 1.0 / 255);
  89.  
  90.             input_tensor_data = Common.ExtractMat(input_img);
  91.  
  92.             input_img.Dispose();
  93.             temp_image.Dispose();
  94.         }
  95.  
  96.         void Postprocess(float[] outputData)
  97.         {
  98.             detectionResults.Clear();
  99.  
  100.             float[] data = Common.Transpose(outputData, class_num + 4, box_num);
  101.  
  102.             float[] confidenceInfo = new float[class_num];
  103.             float[] rectData = new float[4];
  104.  
  105.             List<DetectionResult> detResults = new List<DetectionResult>();
  106.  
  107.             for (int i = 0; i < box_num; i++)
  108.             {
  109.                 Array.Copy(data, i * (class_num + 4), rectData, 0, 4);
  110.                 Array.Copy(data, i * (class_num + 4) + 4, confidenceInfo, 0, class_num);
  111.  
  112.                 float score = confidenceInfo.Max(); // 获取最大值
  113.  
  114.                 int maxIndex = Array.IndexOf(confidenceInfo, score); // 获取最大值的位置
  115.  
  116.                 int _centerX = (int)(rectData[0] * ratio_width);
  117.                 int _centerY = (int)(rectData[1] * ratio_height);
  118.                 int _width = (int)(rectData[2] * ratio_width);
  119.                 int _height = (int)(rectData[3] * ratio_height);
  120.  
  121.                 detResults.Add(new DetectionResult(
  122.                    maxIndex,
  123.                    class_names[maxIndex],
  124.                    new Rect(_centerX - _width / 2, _centerY - _height / 2, _width, _height),
  125.                    score));
  126.             }
  127.  
  128.             //NMS
  129.             CvDnn.NMSBoxes(detResults.Select(x => x.Rect), detResults.Select(x => x.Confidence), conf_threshold, nms_threshold, out int[] indices);
  130.             detResults = detResults.Where((x, index) => indices.Contains(index)).ToList();
  131.  
  132.             detectionResults = detResults;
  133.         }
  134.  
  135.         internal List<DetectionResult> Detect(Mat image)
  136.         {
  137.  
  138.             var t1 = Cv2.GetTickCount();
  139.  
  140.             Stopwatch stopwatch = new Stopwatch();
  141.             stopwatch.Start();
  142.  
  143.             Preprocess(image);
  144.  
  145.             preprocessTime = stopwatch.Elapsed.TotalMilliseconds;
  146.             stopwatch.Restart();
  147.  
  148.             predictor.LoadInferenceData("images", input_tensor_data);
  149.  
  150.             predictor.infer();
  151.  
  152.             inferTime = stopwatch.Elapsed.TotalMilliseconds;
  153.             stopwatch.Restart();
  154.  
  155.             outputData = predictor.GetInferenceResult("output0");
  156.  
  157.             Postprocess(outputData);
  158.  
  159.             postprocessTime = stopwatch.Elapsed.TotalMilliseconds;
  160.             stopwatch.Stop();
  161.  
  162.             totalTime = preprocessTime + inferTime + postprocessTime;
  163.  
  164.             detFps = (double)stopwatch.Elapsed.TotalSeconds / (double)stopwatch.Elapsed.Ticks;
  165.  
  166.             var t2 = Cv2.GetTickCount();
  167.  
  168.             detFps = 1 / ((t2 - t1) / Cv2.GetTickFrequency());
  169.  
  170.             return detectionResults;
  171.  
  172.         }
  173.  
  174.     }
  175. }

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/139265406"。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！