YOLO Training Made Simple: A Unified Pipeline with Mobile Annotation and Web User Interface

We present an open-source, end-to-end object detection training system that integrates mobile annotation, web-based dataset management, and YOLO model training. Users can collect and annotate images via a mobile application. A web interface enables dataset management, training configuration, and online inference. The backend automates data processing, augmentation, training, and inference via API-driven workflows. The system features a modular architecture that supports easy customization and secondary development. It provides a practical, low-barrier pipeline for rapid prototyping, education, and real-world deployment of object detection models. The project is available with MIT license at https://github.com/radiuson/web-yolo.