Application of the Yolov 8 Neural Network for Detection of Personal Protection Equipment in Hazardous Enterprises

The study includes an analysis of the use of artificial neural networks of the YOLO family of architectures for monitoring safety in hazardous enterprises. A software module based on YOLOv8 has been developed for real-time monitoring of personal protective equipment detection in manufacturing. A comparative study of YOLO models was carried out on images obtained from CCTV cameras at the enterprise. YOLOv8 model is due to its high accuracy for image detection and segmentation tasks. (IoU ~90% at 25-30 fps). The trained model recognizes five classes of safety violations. Training was carried out on approximately 1000 images, ensuring effective training and generalization of the model. The software module has expanded authorization and registration functionality, as well as functions for viewing video and analyzing violations with the ability to select the type of violation. A comprehensive solution has been developed for monitoring compliance with safety regulations at industrial facilities. The software includes a desktop application and a chat bot for Telegram. Stack used technologies: Windows Presentation Foundation (WPF), C# and Python languages,.Net Framework, OpenCvSharp.