YoloDotNet 1.1.0

YoloDotNet

YoloDotNet is a C# .NET 8.0 implementation of Yolov8 and ONNX runtime with CUDA

YoloDotNet is a .NET 8 implementation of Yolov8 for detecting objects in images and videos using ML.NET and ONNX runtime with GPU acceleration using CUDA. YoloDotNet currently supports Classification and Object Detection in both images and videos.

Classification	Object Detection
Categorize an image or video frame	Detect multiple objects in a single image or video frame
image from pexels.com	image from pexels.com

Requirements

YoloDotNet with GPU-acceleration requires CUDA and cuDNN.

ℹ️ Before installing CUDA and cuDNN, make sure to verify the ONNX runtime's current compatibility with specific versions.

• Download and install CUDA

• Download cuDNN and follow the installation instructions

• Yolov8 model exported to ONNX format<br> Currently, YoloDotNet supports Classification and ObjectDetection on both images and videos

  Verify your model

  using YoloDotNet;
  
  // Instantiate a new Yolo object with your ONNX-model
  using var yolo = new Yolo(@"path\to\model.onnx");
  
  Console.WriteLine(yolo.OnnxModel.ModelType); // Output if valid: Classification or ObjectDetection
  

[!NOTE] For Video, you need FFmpeg and FFProbe

• Download FFMPEG
• Add FFmpeg and ffprobe to the Path-variable in your Environment Variables

Example - Image Classification

using SixLabors.ImageSharp;
using YoloDotNet;
using YoloDotNet.Extensions;

// Instantiate a new Yolo object with your ONNX-model and CUDA (default)
using var yolo = new Yolo(@"path\to\model_for_classification.onnx");

// Load image
using var image = Image.Load<Rgba32>(@"path\to\image.jpg");

// Run classification, draw labels and save image
var results = yolo.RunClassification(image, 5); // Get top 5 classifications, default = 1
image.DrawClassificationLabels(results);
image.Save(@"path\to\save\image.jpg");

Example - Image Object Detection

using SixLabors.ImageSharp;
using YoloDotNet;
using YoloDotNet.Extensions;

// Instantiate a new Yolo object with your ONNX-model and CUDA (default)
using var yolo = new Yolo(@"path\to\model_for_object_detection.onnx");

// Load image
using var image = Image.Load<Rgba32>(@"path\to\image.jpg");

// Run object detection, draw boxes and save image
var results = yolo.RunObjectDetection(image, 0.5); // Default threshold = 0.25
image.DrawBoundingBoxes(results);
image.Save(@"path\to\save\image.jpg");

Example - Video Classification on frames

using SixLabors.ImageSharp;
using YoloDotNet;
using YoloDotNet.Extensions;

// Instantiate a new Yolo object with your ONNX-model and CUDA
using var yolo = new Yolo(@"path\to\model_for_classification.onnx");

// Run classification and save video to output folder
yolo.RunClassification(new VideoOptions
{
    VideoFile = @"path\to\video.mp4",
    OutputDir = @"path\to\outputfolder"
}, 5); // Get top 5 classifications for each frame, default = 1

Example - Video Object Detection on frames

using SixLabors.ImageSharp;
using YoloDotNet;
using YoloDotNet.Extensions;

// Instantiate a new Yolo object with your ONNX-model and CUDA
using var yolo = new Yolo(@"path\to\model_for_classification.onnx");

// Run object detection and save video to output folder
yolo.RunObjectDetection(new VideoOptions
{
    VideoFile = @"path\to\video.mp4",
    OutputDir = @"path\to\outputfolder"
}, 0.3); // Default threshold = 0.25

GPU

Object detection with GPU and GPU-Id = 0 is enabled by default

// Default setup. GPU with GPU-Id 0
using var yolo = new Yolo(@"path\to\model.onnx");

With a specific GPU-Id

// GPU with a user defined GPU-Id
using var yolo = new Yolo(@"path\to\model.onnx", true, 1);

CPU

YoloDotNet detection with CPU

// With CPU
using var yolo = new Yolo(@"path\to\model.onnx", false);

Access ONNX metadata and labels

The internal ONNX metadata such as input & output parameters, version, author, description, date along with the labels can be accessed via the yolo.OnnxModel property.

Example:

using var yolo = new Yolo(@"path\to\model.onnx");

// ONNX metadata and labels resides inside yolo.OnnxModel
Console.WriteLine(yolo.OnnxModel);

Example:

// Instantiate a new object
using var yolo = new Yolo(@"path\to\model.onnx");

// Display metadata
foreach (var property in yolo.OnnxModel.GetType().GetProperties())
{
    var value = property.GetValue(yolo.OnnxModel);
    Console.WriteLine($"{property.Name,-20}{value!}");

    if (property.Name == nameof(yolo.OnnxModel.CustomMetaData))
        foreach (var data in (Dictionary<string, string>)value!)
            Console.WriteLine($"{"",-20}{data.Key,-20}{data.Value}");
}

// Get ONNX labels
var labels = yolo.OnnxModel.Labels;

Console.WriteLine();
Console.WriteLine($"Labels ({labels.Length}):");
Console.WriteLine(new string('-', 58));

// Display
for (var i = 0; i < labels.Length; i++)
    Console.WriteLine($"index: {i,-8} label: {labels[i].Name,20} color: {labels[i].Color}");

// Output:

// ModelType           ObjectDetection
// InputName           images
// OutputName          output0
// CustomMetaData      System.Collections.Generic.Dictionary`2[System.String,System.String]
//                     date                2023-11-07T13:33:33.565196
//                     description         Ultralytics YOLOv8n model trained on coco.yaml
//                     author              Ultralytics
//                     task                detect
//                     license             AGPL-3.0 https://ultralytics.com/license
//                     version             8.0.202
//                     stride              32
//                     batch               1
//                     imgsz               [640, 640]
//                     names               {0: 'person', 1: 'bicycle', 2: 'car' ... }
// ImageSize           Size [ Width=640, Height=640 ]
// Input               Input { BatchSize = 1, Channels = 3, Width = 640, Height = 640 }
// Output              ObjectDetectionShape { BatchSize = 1, Elements = 84, Channels = 8400 }
// Labels              YoloDotNet.Models.LabelModel[]
//
// Labels (80):
// ---------------------------------------------------------
// index: 0        label: person              color: #5d8aa8
// index: 1        label: bicycle             color: #f0f8ff
// index: 2        label: car                 color: #e32636
// index: 3        label: motorcycle          color: #efdecd
// ...

Donate

https://paypal.me/nickswardh

References & Acknowledgements

https://github.com/ultralytics/ultralytics

https://github.com/sstainba/Yolov8.Net

https://github.com/mentalstack/yolov5-net