CSharp-Yolo-Video

Although the C# wrapper for Darknet exists, I went through a hard time figuring out how to apply the wrapper for videos. For later use for myself and saving others' time, I summarize how to apply the Yolo wrapper on videos.

Getting Started

The following instructions will lead you to setting the environment for using the Yolo wrapper in your project.

System requriements

• .NET Framework 4.6.1
• Microsoft Visual C++ Redistributable for Visual Studio 2017 x64

Install Alturos.Yolo using NuGet as follows or you can use the NuGet GUI instead. If you're using NuGet GUI, search for "Alturos.Yolo".

PM> install-package Alturos.Yolo (C# wrapper and C++ dlls 28MB)
PM> install-package Alturos.YoloV2TinyVocData (YOLOv2-tiny Pre-Trained Dataset 56MB)

(Optional) For GPU support, install and download the followings.

1. Install the latest Nvidia driver for your graphic device.
2. Install Nvidia CUDA Toolkit 10.1 (must be installed add a hardware driver for cuda support)
3. Download Nvidia cuDNN v7.6.3 for CUDA 10.1
4. Copy the cudnn64_7.dll from the output directory of cdDNN v7.6.3. into the x64 folder of your project.

Install OpenCvSharp3-AnyCPU over NuGet as follows or search for "OpenCvSharp3-AnyCPU". Although the package name contains CvSharp3, it is actually an OpenCv 4.x wrapper.

PM> install-package OpenCvSharp3-AnyCPU

Download pretrained weights and place it in your project directory. For more information, visit Alturos.Yolo

Model	Processing Resolution	Cfg	Weights	Names
YOLOv3	608x608	yolov3.cfg	yolov3.weights	coco.names
YOLOv3-tiny	416x416	yolov3-tiny.cfg	yolov3-tiny.weights	coco.names
YOLOv2	608x608	yolov2.cfg	yolov2.weights	coco.names
YOLOv2-tiny	416x416	yolov2-tiny.cfg	yolov2-tiny.weights	voc.names
yolo9000	448x448	yolo9000.cfg	yolo9000.weights	9k.names

Write Codes for Video Object Recognition

The following is the minimum code for running the Yolo wrapper on a video file. For running the code, set the solution platform as "x64"!

using OpenCvSharp;
using OpenCvSharp.Extensions;

using Alturos.Yolo;

private void VideoObjectDetection()
{
  // YOLO setting
  int yoloWidth = 608, yoloHeight = 608;
  var configurationDetector = new ConfigurationDetector();
  var config = configurationDetector.Detect();
  YoloWrapper yoloWrapper = new YoloWrapper(config);
  
  // OpenCV & WPF setting
  VideoCapture videocapture;
  Mat image = new Mat();
  WriteableBitmap wb = new WriteableBitmap(yoloWidth, yoloHeight, 96, 96, PixelFormats.Bgr24, null);
  
  byte[] imageInBytes = new byte[(int)(yoloWidth * yoloHeight * image.Channels())];
  
  // Read a video file and run object detection over it!
  using (videocapture = new VideoCapture(address))
  {
    using(Mat imageOriginal = new Mat())
    {
      // read a single frame and convert the frame into a byte array
      videocapture.Read(imageOriginal);
      image = imageOriginal.Resize(new OpenCvSharp.Size(yoloWidth, yoloHeight));
      imageInBytes = image.ToBytes();
      
      // conduct object detection and display the result
      var items = yolowrapper.Detect(imageInBytes);
      foreach(var item in items)
      {
        var x = item.X;
        var y = item.Y;
        var width = item.Width;
        var height = item.Height;
        var type = item.Type;  // class name of the object
        
        // draw a bounding box for the detected object
        // you can set different colors for different classes
        Cv2.Rectangle(image, new OpenCvSharp.Rect(x, y, width, height), Scalar.Green, 3);
      }
      
      // display the detection result
      WriteableBitmapConverter.ToWriteableBitmap(image, wb);
      /* WPF component: videoViewer
      <Canvas Name="canvasYoloVideo" Height="608" Width="608">
        <Image Name="videoViewer" Height="608" Width="608" Stretch="Fill" />
      </Canvas>
      */
      videoViewer.Source = wb;
    }
  }
}