代码语言：javascript复制
using System;
using System.Collections.Generic;
using System.Diagnostics;
using System.Drawing;
using System.IO;
using System.Runtime.ExceptionServices;
using System.Runtime.InteropServices;
using System.Security;
using System.Threading;
using System.Threading.Tasks;
using System.Windows.Forms;
using Emgu.CV;
using Emgu.CV.CvEnum;
using Emgu.CV.Structure;
namespace WindowsFormsApp1
{
public partial class Form1 : Form
{
private const string YoloLibraryName = "yolo_cpp_dll.dll";
private const int MaxObjects = 1000;
object ThreadLock = new object();
[DllImport(YoloLibraryName, EntryPoint = "init")]
private static extern int InitializeYolo(string configurationFilename, string weightsFilename, int gpu);
[DllImport(YoloLibraryName, EntryPoint = "detect_image")]
private static extern int DetectImage(string filename, ref BboxContainer container);
[DllImport(YoloLibraryName, EntryPoint = "detect_mat")]
//private static extern int DetectImage(IntPtr pArray, int nSize, ref BboxContainer container);
private static extern int DetectImage(uint width, uint height, IntPtr pArray, int nSize, ref BboxContainer container);
[DllImport(YoloLibraryName, EntryPoint = "dispose")]
private static extern int DisposeYolo();
[StructLayout(LayoutKind.Sequential)]
public struct BboxContainer
{
[MarshalAs(UnmanagedType.ByValArray, SizeConst = MaxObjects)]
public bbox_t[] candidates;
}
[StructLayout(LayoutKind.Sequential)]
public struct bbox_t
{
public UInt32 x, y, w, h; // (x,y) - top-left corner, (w, h) - width & height of bounded box
public float prob; // confidence - probability that the object was found correctly
public UInt32 obj_id; // class of object - from range [0, classes-1]
public UInt32 track_id; // tracking id for video (0 - untracked, 1 - inf - tracked object)
public UInt32 frames_counter;
float x_3d, y_3d, z_3d; // center of object (in Meters) if ZED 3D Camera is used 必须和yolo_cpp_dll.dll中的定义一一对应
};
public Form1()
{
InitializeComponent();
//CheckForIllegalCrossThreadCalls = false;
}
private void Form1_Load(object sender, EventArgs e)
{
//Task.Run(() => InitYolo());
InitYolo();
}
public static void InitYolo()
{
InitializeYolo(
Path.Combine(AppDomain.CurrentDomain.BaseDirectory, "yolov3.cfg"),
Path.Combine(AppDomain.CurrentDomain.BaseDirectory, "yolov3.weights"),
0);
}
[HandleProcessCorruptedStateExceptions]
[SecurityCritical]
public List<bbox_t> Detect(int Width, int Height, byte[] imageData)
{
var container = new BboxContainer();
var size = Marshal.SizeOf(imageData[0]) * imageData.Length;
var pnt = Marshal.AllocHGlobal(size);
try
{
Marshal.Copy(imageData, 0, pnt, imageData.Length);
var count = DetectImage((uint)Width, (uint)Height, pnt, imageData.Length, ref container);
if (count == -1)
{
throw new NotSupportedException(" has no OpenCV support");
}
List<bbox_t> result = new List<bbox_t>();
for (int i = 0; i < count; i )
{
result.Add(container.candidates[i]);
}
return result;
}
catch (Exception exception)
{
Console.WriteLine(exception.Message);
return new List<bbox_t>();
}
finally
{
// Free the unmanaged memory.
Marshal.FreeHGlobal(pnt); //不释放内存会报错
}
}
private void button1_Click(object sender, EventArgs e)
{
string[] str = new string[10] { "Camera20200312154951725.jpg" , "Camera20200312155912410.jpg", "Camera20200312155923659.jpg",
"Camera20200312162258694.jpg","Camera20200313100654338.jpg","Camera20200313155526143.jpg",
"Camera20200313161231227.jpg","Camera20200319102951733.jpg","Camera20200319111317754.jpg",
"Camera20200320135659163.jpg"};
long timeCount = 0;
object locker = new object();
for (int i = 0; i < 10; i )
{
//Image<Bgr, byte> img = new Image<Bgr, byte>(@"C:UsersadminsourcereposWindowsFormsApp1WindowsFormsApp1binx64DebugCamera20200421194241118.jpg");
Image<Bgr, byte> img1 = new Image<Bgr, byte>(@"C:UsersadminDesktop 512" str[i]);
//Image<Gray, byte> img1 = new Image<Gray, byte>(@"C:UsersadminsourcereposWindowsFormsApp1WindowsFormsApp1binx64DebugDevice20200426191227506.jpg");
//Mat img2 = CvInvoke.Imread(@"C:UsersadminsourcereposWindowsFormsApp1WindowsFormsApp1binx64DebugCamera20200421194241118.jpg", Emgu.CV.CvEnum.LoadImageType.Grayscale);
int Width, Height;
Width = img1.Width;
Height = img1.Height;
while (Width % 4 != 0)
{
Width ;
}
CvInvoke.Resize(img1, img1, new Size(Width, Height), 0, 0, Inter.Lanczos4);
//Matrix<byte> showImage = new Matrix<byte>(img1.Height, img1.Width, 3);
//Matrix<byte> showImage = new Matrix<byte>(img1.Height, img1.Width, img1.Bytes[img1.Height* img1.Width*]);
//CvInvoke.CvtColor(img1, showImage, ColorConversion.Gray2Bgr);
List<bbox_t> bboxes = new List<bbox_t>();
lock (ThreadLock) //锁线程
{
Stopwatch sw = new Stopwatch();
sw.Start();
bboxes = Detect(img1.Height, img1.Width, img1.Bytes);
sw.Stop();
//MessageBox.Show(sw.ElapsedMilliseconds.ToString() "毫秒");
timeCount = sw.ElapsedMilliseconds;
}
if (bboxes.ToArray().Length != 0)
{
//MessageBox.Show("1111");
}
foreach (var bbox in bboxes)
{
//detectobject newobj = new detectobject();
var color_red = new MCvScalar(0, 0, 255); // BGR
var color_green = new MCvScalar(0, 255, 0);
var color_yellow = new MCvScalar(0, 255, 255);
Rectangle rect = new Rectangle((int)bbox.x, (int)bbox.y, (int)bbox.w, (int)bbox.h);
if (bbox.obj_id == 0)
{
// IsWarning = true;
CvInvoke.Rectangle(img1, rect, color_yellow);
//string hiddenMsg = labels[(int)bbox.obj_id] "," bbox.prob.ToString("0.00");
//CvInvoke.PutText(showImage, hiddenMsg, rect.Location, FontFace.HersheyComplex, 0.8, color_red);
}
else if (bbox.obj_id == 1)
{
//IsWarning = true;
CvInvoke.Rectangle(img1, rect, color_green);
//string hiddenMsg = labels[(int)bbox.obj_id] "," bbox.prob.ToString("0.00");
//CvInvoke.PutText(showImage, hiddenMsg, rect.Location, FontFace.HersheyComplex, 0.8, color_green);
}
else
{
//IsWarning = true;
CvInvoke.Rectangle(img1, rect, color_red);
//string hiddenMsg = labels[(int)bbox.obj_id] "," bbox.prob.ToString("0.00");
//CvInvoke.PutText(showImage, hiddenMsg, rect.Location, FontFace.HersheyComplex, 0.8, color_green);
}
}
Task.Run(() =>
{
lock (locker)
{
Thread.Sleep(1000);
pictureBox1.Image = img1.Mat.Bitmap;
//this.BeginInvoke(new Action(() => { pictureBox1.Image = showImage.Mat.Bitmap; }));
}
});
}
label1.Text = (timeCount / 10).ToString() "毫秒";
}
}
}
结果：
结论：
测试10张图片，计算平均耗时： yolo_cpp_dll.dll 自己编译生成的 yolov3 18ms 100% yolov3-tiny 5ms 检出率60% 比较节省gup资源 yolov4 35ms 检出率90%
yolo_cpp_dll_gpu.dll--旧版的dll，以前同事留下的 yolov3 19ms 70% yolov3-tiny 6ms 检出率40% yolov4 34ms 检出率0%
结论：dll对测试效果也有影响，yolov3-tiny检出速度快但是检出率不高，yolov4虽然做了优化，但是在耗时和检出率上反而下降了 ，原因不详！
