Computer Vision Method for Detecting Safety Violations Using the YOLOv8 Model

In today's world, where production processes are becoming increasingly automated, ensuring safety on production sites is becoming a priority. Conventional methods of monitoring and enforcing safety regulations require large resources and are not always effective. However, with the development of machine learning technologies, new prospects are emerging in the field of detecting safety violations in production. An information system has been developed for monitoring and analyzing video from cameras, which automatically detects violations and generates appropriate reports. To train the system, a neural network based on the YOLOv8 model was used. The automatic parameter tuning algorithm was implemented using the gradient descent method to optimize learning.