Využití kamerového systému pro zajištěni bezpečnosti osob na pracovišti

Abstract

The development of technology in recent years, particularly in the field of artificial intelligence and neural networks, enables more complex analyses of human behavior. Security cameras could thus be used to ensure personal safety. This work proposes a solution for real-time detection of human falls in video streams. The solution is based on a combination of two neural networks. The first detects all people in the image and their key points, the second classifies these points into the categories "normal" and "fallen". The work describes the selection of a pose detector, the design of the classification network architecture, and the implementation of the resulting fall detector. The final solution uses the YOLOv11-pose model and a recurrent neural network based on the GRU architecture.