Practical overview
When it fits
Use it when a .NET application needs to run supported ONNX-based vision models on images or live video, with the inference integrated in C# rather than through a Python runtime.
Who it is for
- C# developers building image or video computer-vision features into .NET applications.
- Teams developing desktop applications, backend services or real-time vision pipelines.
Documented uses
- Run image classification or object detection with an ONNX model.
- Process live video for detection, segmentation or pose estimation.
- Add oriented bounding-box detection to a .NET computer-vision workflow.
- Choose an execution provider and configure preprocessing and inference thresholds.
Requirements & limits
The project is a public GitHub repository and is MIT licensed. Its README identifies .NET 8 as the foundation and lists Windows, Linux and macOS as supported platforms.
- The library uses ONNX models; export opset requirements vary by model family.
- The README directs users to install exactly one execution-provider package.
- Preprocessing and confidence/IoU settings need to match the model’s training configuration.
- The library’s MIT license is separate from model licenses; check the license for the selected model.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Run image object detection with the CPU provider
Use the README quick start to run object detection on an image and save an annotated result.
Before you start
- A .NET 8 project.
- The YoloDotNet core package and CPU execution-provider package.
- An ONNX model file and an image file.
Steps
Install the core package with
dotnet add package YoloDotNet.Install the CPU provider with
dotnet add package YoloDotNet.ExecutionProvider.Cpu.In C#, create a Yolo instance with a CpuExecutionProvider pointing to the ONNX model, then decode the image with SkiaSharp.
Call
RunObjectDetection, draw the returned results on the image, and save the output image.
Expected outcome
The example produces object-detection results and saves an image with the results drawn on it.
Guide limitations
- The example uses a CPU provider; other providers are separate package choices, and the README says to install exactly one.
- The README says preprocessing and confidence/IoU settings should match the model’s training configuration.
- For YOLOv5u–YOLOv12, export with opset 17; for YOLOv26, use opset 18. The documented example notes IoU is ignored for YOLOv10 and YOLOv26.
- YoloDotNet is MIT licensed, but model licensing is separate; most Ultralytics YOLO models are AGPL-3.0 or require a commercial license for commercial use.
Compare options for Model inference and serving
These projects document a common task and share a project role. This is a source-based comparison, not an execution benchmark or a guarantee of interchangeability.
| Project | Documented purpose | Platform | Access and requirements |
|---|---|---|---|
| Yolodotnet | Run image classification or object detection with an ONNX model. | Linux, macOS, Server, Windows | The project is a public GitHub repository and is MIT licensed. Its README identifies .NET 8 as the foundation and lists Windows, Linux and macOS as supported platforms. |
| Hyperpose | Customize pose-estimation architectures and datasets. | Docker, Linux, macOS, Windows | |
| Multi Task Nlp | Define multiple tasks and their data and model settings in a YAML file. | Linux, macOS, Windows | The project is an open-source library under the Apache-2.0 license, distributed through its public GitHub repository. |
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 14 October 2026
- Original website title
- NickSwardh/YoloDotNet · Official homepage title · retained in its source language