Person Detection for Monitoring Individuals Accessing the Robot Working Zones Using YOLOv8

In active workshops, methods to warn of dangerous areas are essential to ensure worker safety. However, traditional methods such as having guards, posting warning signs, etc. are not enough. With the strong development of object-tracking technologies, algorithms have achieved high accuracy and are applied in practice. This study proposes the YOLOv8 algorithm to monitor human posture in robot working zones. We use the Shapely library to create warning zones and check for intruders. Finally, the python-telegram-bot library sends a warning about Telegram. The system is tested and evaluated in normal and low-light environmental conditions, with one person or many people. The model achieves good performance, responding in real-time.