C# OpenVINO Yolov8-OBB 旋转目标检测

目录

效果

模型

项目

代码

下载 

C# OpenVINO Yolov8-OBB 旋转目标检测

效果

模型

Model Properties
-------------------------
date：2024-02-26T08:38:44.171849
description：Ultralytics YOLOv8s-obb model trained on runs/DOTAv1.0-ms.yaml
author：Ultralytics
task：obb
license：AGPL-3.0 https://ultralytics.com/license
version：8.1.18
stride：32
batch：1
imgsz：[640, 640]
names：{0: 'plane', 1: 'ship', 2: 'storage tank', 3: 'baseball diamond', 4: 'tennis court', 5: 'basketball court', 6: 'ground track field', 7: 'harbor', 8: 'bridge', 9: 'large vehicle', 10: 'small vehicle', 11: 'helicopter', 12: 'roundabout', 13: 'soccer ball field', 14: 'swimming pool'}
---------------------------------------------------------------

Inputs
-------------------------
name：images
tensor：Float[1, 3, 640, 640]
---------------------------------------------------------------

Outputs
-------------------------
name：output0
tensor：Float[1, 20, 8400]
---------------------------------------------------------------

项目

代码

using OpenCvSharp;
using OpenCvSharp.Dnn;
using Sdcb.OpenVINO;
using Sdcb.OpenVINO.Natives;
using System;
using System.Collections.Generic;
using System.Diagnostics;
using System.Drawing;
using System.IO;
using System.Text;
using System.Windows.Forms;

namespace OpenVINO_Det
{
    public partial class Form1 : Form
    {
        public Form1()
        {
            InitializeComponent();
        }

        string fileFilter = "*.*|*.bmp;*.jpg;*.jpeg;*.tiff;*.tiff;*.png";
        string image_path = "";
        string classer_path;
        string model_path;
        Mat image;
        Mat result_image;

        string[] class_lables;

        StringBuilder sb = new StringBuilder();

        private void button1_Click(object sender, EventArgs e)
        {
            OpenFileDialog ofd = new OpenFileDialog();
            ofd.Filter = fileFilter;
            if (ofd.ShowDialog() != DialogResult.OK) return;
            pictureBox1.Image = null;
            image_path = ofd.FileName;
            pictureBox1.Image = new Bitmap(image_path);
            textBox1.Text = "";
            pictureBox2.Image = null;
        }

        unsafe private void button2_Click(object sender, EventArgs e)
        {
            if (pictureBox1.Image == null)
            {
                return;
            }

            pictureBox2.Image = null;
            textBox1.Text = "";
            sb.Clear();
            Application.DoEvents();

            Model rawModel = OVCore.Shared.ReadModel(model_path);
            PrePostProcessor pp = rawModel.CreatePrePostProcessor();
            PreProcessInputInfo inputInfo = pp.Inputs.Primary;

            inputInfo.TensorInfo.Layout = Sdcb.OpenVINO.Layout.NHWC;
            inputInfo.ModelInfo.Layout = Sdcb.OpenVINO.Layout.NCHW;

            Model m = pp.BuildModel();
            CompiledModel cm = OVCore.Shared.CompileModel(m, "CPU");
            InferRequest ir = cm.CreateInferRequest();
           
            Stopwatch stopwatch = new Stopwatch();

            //图片缩放
            image = new Mat(image_path);
            int max_image_length = image.Cols > image.Rows ? image.Cols : image.Rows;
            Mat max_image = Mat.Zeros(new OpenCvSharp.Size(max_image_length, max_image_length), MatType.CV_8UC3);
            Rect roi = new Rect(0, 0, image.Cols, image.Rows);
            image.CopyTo(new Mat(max_image, roi));

            float factor = (float)(max_image_length / 640.0);

            // 将图片转为RGB通道
            Mat image_rgb = new Mat();
            Cv2.CvtColor(max_image, image_rgb, ColorConversionCodes.BGR2RGB);
            Mat resize_image = new Mat();
            Cv2.Resize(image_rgb, resize_image, new OpenCvSharp.Size(640, 640));

            Mat f32 = new Mat();
            resize_image.ConvertTo(f32, MatType.CV_32FC3, 1.0 / 255);

            using (Tensor input = Tensor.FromRaw(
                 new ReadOnlySpan((void*)f32.Data, (int)((long)f32.DataEnd - (long)f32.DataStart)),
                new Shape(1, f32.Rows, f32.Cols, 3),
                ov_element_type_e.F32))
            {
                ir.Inputs.Primary = input;
            }
            double preprocessTime = stopwatch.Elapsed.TotalMilliseconds;
            stopwatch.Restart();

            ir.Run();
            double inferTime = stopwatch.Elapsed.TotalMilliseconds;
            stopwatch.Restart();

            using (Tensor output = ir.Outputs.Primary)
            {
                ReadOnlySpan data = output.GetData();

                Mat result_data = new Mat(20, 8400, MatType.CV_32F, data.ToArray());
                result_data = result_data.T();
                List position_boxes = new List();
                List class_ids = new List();
                List confidences = new List();
                List rotations = new List();
                // Preprocessing output results
                for (int i = 0; i < result_data.Rows; i++)
                {
                    Mat classes_scores = new Mat(result_data, new Rect(4, i, 15, 1));
                    OpenCvSharp.Point max_classId_point, min_classId_point;
                    double max_score, min_score;
                    // Obtain the maximum value and its position in a set of data
                    Cv2.MinMaxLoc(classes_scores, out min_score, out max_score,
                        out min_classId_point, out max_classId_point);
                    // Confidence level between 0 ~ 1
                    // Obtain identification box information
                    if (max_score > 0.25)
                    {
                        float cx = result_data.At(i, 0);
                        float cy = result_data.At(i, 1);
                        float ow = result_data.At(i, 2);
                        float oh = result_data.At(i, 3);
                        double x = (cx - 0.5 * ow) * factor;
                        double y = (cy - 0.5 * oh) * factor;
                        double width = ow * factor;
                        double height = oh * factor;
                        Rect2d box = new Rect2d();
                        box.X = x;
                        box.Y = y;
                        box.Width = width;
                        box.Height = height;
                        position_boxes.Add(box);
                        class_ids.Add(max_classId_point.X);
                        confidences.Add((float)max_score);
                        rotations.Add(result_data.At(i, 19));
                    }
                }

                // NMS 
                int[] indexes = new int[position_boxes.Count];
                CvDnn.NMSBoxes(position_boxes, confidences, 0.25f, 0.7f, out indexes);
                List rotated_rects = new List();
                for (int i = 0; i < indexes.Length; i++)
                {
                    int index = indexes[i];
                    float w = (float)position_boxes[index].Width;
                    float h = (float)position_boxes[index].Height;
                    float x = (float)position_boxes[index].X + w / 2;
                    float y = (float)position_boxes[index].Y + h / 2;
                    float r = rotations[index];
                    float w_ = w > h ? w : h;
                    float h_ = w > h ? h : w;
                    r = (float)((w > h ? r : (float)(r + Math.PI / 2)) % Math.PI);
                    RotatedRect rotate = new RotatedRect(new Point2f(x, y), new Size2f(w_, h_), (float)(r * 180.0 / Math.PI));
                    rotated_rects.Add(rotate);
                }

                double postprocessTime = stopwatch.Elapsed.TotalMilliseconds;
                stopwatch.Stop();
                double totalTime = preprocessTime + inferTime + postprocessTime;

                result_image = image.Clone();

                for (int i = 0; i < indexes.Length; i++)
                {
                    int index = indexes[i];
                    Point2f[] points = rotated_rects[i].Points();

                    for (int j = 0; j < 4; j++)
                    {
                        Cv2.Line(result_image, (OpenCvSharp.Point)points[j], (OpenCvSharp.Point)points[(j + 1) % 4], new Scalar(0, 255, 0), 2);
                    }

                    Cv2.PutText(result_image, class_lables[class_ids[index]] + "-" + confidences[index].ToString("0.00"),
                        (OpenCvSharp.Point)points[0], HersheyFonts.HersheySimplex, 0.8, new Scalar(0, 0, 255), 2);
                }

                sb.AppendLine($"Preprocess: {preprocessTime:F2}ms");
                sb.AppendLine($"Infer: {inferTime:F2}ms");
                sb.AppendLine($"Postprocess: {postprocessTime:F2}ms");
                sb.AppendLine($"Total: {totalTime:F2}ms");

                pictureBox2.Image = new Bitmap(result_image.ToMemoryStream());
                textBox1.Text = sb.ToString();
            }
        }

        private void Form1_Load(object sender, EventArgs e)
        {
            model_path = "yolov8s-obb.onnx";
            classer_path = "lable.txt";

            List str = new List();
            StreamReader sr = new StreamReader(classer_path);
            string line;
            while ((line = sr.ReadLine()) != null)
            {
                str.Add(line);
            }
            class_lables = str.ToArray();

            image_path = "2.png";
            pictureBox1.Image = new Bitmap(image_path);
        }
    }
}

  1. using OpenCvSharp;
  2. using OpenCvSharp.Dnn;
  3. using Sdcb.OpenVINO;
  4. using Sdcb.OpenVINO.Natives;
  5. using System;
  6. using System.Collections.Generic;
  7. using System.Diagnostics;
  8. using System.Drawing;
  9. using System.IO;
  10. using System.Text;
  11. using System.Windows.Forms;
  12.  
  13. namespace OpenVINO_Det
  14. {
  15.     public partial class Form1 : Form
  16.     {
  17.         public Form1()
  18.         {
  19.             InitializeComponent();
  20.         }
  21.  
  22.         string fileFilter = "*.*|*.bmp;*.jpg;*.jpeg;*.tiff;*.tiff;*.png";
  23.         string image_path = "";
  24.         string classer_path;
  25.         string model_path;
  26.         Mat image;
  27.         Mat result_image;
  28.  
  29.         string[] class_lables;
  30.  
  31.         StringBuilder sb = new StringBuilder();
  32.  
  33.         private void button1_Click(object sender, EventArgs e)
  34.         {
  35.             OpenFileDialog ofd = new OpenFileDialog();
  36.             ofd.Filter = fileFilter;
  37.             if (ofd.ShowDialog() != DialogResult.OK) return;
  38.             pictureBox1.Image = null;
  39.             image_path = ofd.FileName;
  40.             pictureBox1.Image = new Bitmap(image_path);
  41.             textBox1.Text = "";
  42.             pictureBox2.Image = null;
  43.         }
  44.  
  45.         unsafe private void button2_Click(object sender, EventArgs e)
  46.         {
  47.             if (pictureBox1.Image == null)
  48.             {
  49.                 return;
  50.             }
  51.  
  52.             pictureBox2.Image = null;
  53.             textBox1.Text = "";
  54.             sb.Clear();
  55.             Application.DoEvents();
  56.  
  57.             Model rawModel = OVCore.Shared.ReadModel(model_path);
  58.             PrePostProcessor pp = rawModel.CreatePrePostProcessor();
  59.             PreProcessInputInfo inputInfo = pp.Inputs.Primary;
  60.  
  61.             inputInfo.TensorInfo.Layout = Sdcb.OpenVINO.Layout.NHWC;
  62.             inputInfo.ModelInfo.Layout = Sdcb.OpenVINO.Layout.NCHW;
  63.  
  64.             Model m = pp.BuildModel();
  65.             CompiledModel cm = OVCore.Shared.CompileModel(m, "CPU");
  66.             InferRequest ir = cm.CreateInferRequest();
  67.            
  68.             Stopwatch stopwatch = new Stopwatch();
  69.  
  70.             //图片缩放
  71.             image = new Mat(image_path);
  72.             int max_image_length = image.Cols > image.Rows ? image.Cols : image.Rows;
  73.             Mat max_image = Mat.Zeros(new OpenCvSharp.Size(max_image_length, max_image_length), MatType.CV_8UC3);
  74.             Rect roi = new Rect(0, 0, image.Cols, image.Rows);
  75.             image.CopyTo(new Mat(max_image, roi));
  76.  
  77.             float factor = (float)(max_image_length / 640.0);
  78.  
  79.             // 将图片转为RGB通道
  80.             Mat image_rgb = new Mat();
  81.             Cv2.CvtColor(max_image, image_rgb, ColorConversionCodes.BGR2RGB);
  82.             Mat resize_image = new Mat();
  83.             Cv2.Resize(image_rgb, resize_image, new OpenCvSharp.Size(640, 640));
  84.  
  85.             Mat f32 = new Mat();
  86.             resize_image.ConvertTo(f32, MatType.CV_32FC3, 1.0 / 255);
  87.  
  88.             using (Tensor input = Tensor.FromRaw(
  89.                  new ReadOnlySpan<byte>((void*)f32.Data, (int)((long)f32.DataEnd - (long)f32.DataStart)),
  90.                 new Shape(1, f32.Rows, f32.Cols, 3),
  91.                 ov_element_type_e.F32))
  92.             {
  93.                 ir.Inputs.Primary = input;
  94.             }
  95.             double preprocessTime = stopwatch.Elapsed.TotalMilliseconds;
  96.             stopwatch.Restart();
  97.  
  98.             ir.Run();
  99.             double inferTime = stopwatch.Elapsed.TotalMilliseconds;
  100.             stopwatch.Restart();
  101.  
  102.             using (Tensor output = ir.Outputs.Primary)
  103.             {
  104.                 ReadOnlySpan<float> data = output.GetData<float>();
  105.  
  106.                 Mat result_data = new Mat(20, 8400, MatType.CV_32F, data.ToArray());
  107.                 result_data = result_data.T();
  108.                 List<Rect2d> position_boxes = new List<Rect2d>();
  109.                 List<int> class_ids = new List<int>();
  110.                 List<float> confidences = new List<float>();
  111.                 List<float> rotations = new List<float>();
  112.                 // Preprocessing output results
  113.                 for (int i = 0; i < result_data.Rows; i++)
  114.                 {
  115.                     Mat classes_scores = new Mat(result_data, new Rect(4, i, 15, 1));
  116.                     OpenCvSharp.Point max_classId_point, min_classId_point;
  117.                     double max_score, min_score;
  118.                     // Obtain the maximum value and its position in a set of data
  119.                     Cv2.MinMaxLoc(classes_scores, out min_score, out max_score,
  120.                         out min_classId_point, out max_classId_point);
  121.                     // Confidence level between 0 ~ 1
  122.                     // Obtain identification box information
  123.                     if (max_score > 0.25)
  124.                     {
  125.                         float cx = result_data.At<float>(i, 0);
  126.                         float cy = result_data.At<float>(i, 1);
  127.                         float ow = result_data.At<float>(i, 2);
  128.                         float oh = result_data.At<float>(i, 3);
  129.                         double x = (cx - 0.5 * ow) * factor;
  130.                         double y = (cy - 0.5 * oh) * factor;
  131.                         double width = ow * factor;
  132.                         double height = oh * factor;
  133.                         Rect2d box = new Rect2d();
  134.                         box.X = x;
  135.                         box.Y = y;
  136.                         box.Width = width;
  137.                         box.Height = height;
  138.                         position_boxes.Add(box);
  139.                         class_ids.Add(max_classId_point.X);
  140.                         confidences.Add((float)max_score);
  141.                         rotations.Add(result_data.At<float>(i, 19));
  142.                     }
  143.                 }
  144.  
  145.                 // NMS 
  146.                 int[] indexes = new int[position_boxes.Count];
  147.                 CvDnn.NMSBoxes(position_boxes, confidences, 0.25f, 0.7f, out indexes);
  148.                 List<RotatedRect> rotated_rects = new List<RotatedRect>();
  149.                 for (int i = 0; i < indexes.Length; i++)
  150.                 {
  151.                     int index = indexes[i];
  152.                     float w = (float)position_boxes[index].Width;
  153.                     float h = (float)position_boxes[index].Height;
  154.                     float x = (float)position_boxes[index].X + w / 2;
  155.                     float y = (float)position_boxes[index].Y + h / 2;
  156.                     float r = rotations[index];
  157.                     float w_ = w > h ? w : h;
  158.                     float h_ = w > h ? h : w;
  159.                     r = (float)((w > h ? r : (float)(r + Math.PI / 2)) % Math.PI);
  160.                     RotatedRect rotate = new RotatedRect(new Point2f(x, y), new Size2f(w_, h_), (float)(r * 180.0 / Math.PI));
  161.                     rotated_rects.Add(rotate);
  162.                 }
  163.  
  164.                 double postprocessTime = stopwatch.Elapsed.TotalMilliseconds;
  165.                 stopwatch.Stop();
  166.                 double totalTime = preprocessTime + inferTime + postprocessTime;
  167.  
  168.                 result_image = image.Clone();
  169.  
  170.                 for (int i = 0; i < indexes.Length; i++)
  171.                 {
  172.                     int index = indexes[i];
  173.                     Point2f[] points = rotated_rects[i].Points();
  174.  
  175.                     for (int j = 0; j < 4; j++)
  176.                     {
  177.                         Cv2.Line(result_image, (OpenCvSharp.Point)points[j], (OpenCvSharp.Point)points[(j + 1) % 4], new Scalar(0, 255, 0), 2);
  178.                     }
  179.  
  180.                     Cv2.PutText(result_image, class_lables[class_ids[index]] + "-" + confidences[index].ToString("0.00"),
  181.                         (OpenCvSharp.Point)points[0], HersheyFonts.HersheySimplex, 0.8, new Scalar(0, 0, 255), 2);
  182.                 }
  183.  
  184.                 sb.AppendLine($"Preprocess: {preprocessTime:F2}ms");
  185.                 sb.AppendLine($"Infer: {inferTime:F2}ms");
  186.                 sb.AppendLine($"Postprocess: {postprocessTime:F2}ms");
  187.                 sb.AppendLine($"Total: {totalTime:F2}ms");
  188.  
  189.                 pictureBox2.Image = new Bitmap(result_image.ToMemoryStream());
  190.                 textBox1.Text = sb.ToString();
  191.             }
  192.         }
  193.  
  194.         private void Form1_Load(object sender, EventArgs e)
  195.         {
  196.             model_path = "yolov8s-obb.onnx";
  197.             classer_path = "lable.txt";
  198.  
  199.             List<string> str = new List<string>();
  200.             StreamReader sr = new StreamReader(classer_path);
  201.             string line;
  202.             while ((line = sr.ReadLine()) != null)
  203.             {
  204.                 str.Add(line);
  205.             }
  206.             class_lables = str.ToArray();
  207.  
  208.             image_path = "2.png";
  209.             pictureBox1.Image = new Bitmap(image_path);
  210.         }
  211.     }
  212. }

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/136327087"。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！