YoloDotNet

Why we included this project

C# teams that need YOLO inference usually end up bolting a Python service onto the side of their stack. YoloDotNet removes that step: it is a pure .NET 8 library built on ONNX Runtime and SkiaSharp, with no OpenCV dependency and no Python runtime, so behavior stays explicit and predictable. It handles classification, object detection, oriented bounding boxes, segmentation, and pose estimation in both still images and live video, and runs on CPU as well as CUDA, TensorRT, OpenVINO, CoreML, and DirectML. Recent additions such as region-of-interest inference, JSON export, and YOLO-format annotation output make it handy for surveillance zones, monitoring pipelines, and generating training data from your own code. For .NET developers who want production-grade YOLO without leaving the ecosystem, it is a solid choice.