Adaptive Zone Monitoring for Industrial Safety: Integrating YOLO with Pan Tilt Zoom Camera Control and Human Machine Interfaces

Document Type

Conference Proceeding

Publication Date

9-21-2026

Abstract

Industrial environments necessitate robust safety systems to prevent accidents involving human machine interaction. Traditional safety measures like physical barriers and light curtains often lack the flexibility required for dynamic modern workplaces. This paper presents a vision based safety zone monitoring system that integrates a Pan, Tilt, Zoom (PTZ) camera, deep learning based object detection, and industrial communication protocols. The system employs YOLOv8 for realtime human detection, with a Human Machine Interface (HMI) that enables zone creation, PTZ control, database management, and multilanguage support. Safety logic is implemented by continuously evaluating whether detected persons intersect user-defined polygonal zones, triggering a binary signal when a restricted area is breached. Communication with external devices is achieved through three channels: Allen-Bradley CompactLogix 1769-L16ER PLCs using the pycomm3 library, configurable serial connections, and MQTT messaging. The system was deployed in a real industrial setup using an Axis P5655-E PTZ camera (1080p), where it demonstrated reliable operation. Experimental results show an average inference time of less than 52 ms per frame on an RTX 5070 Ti GPU, corresponding to approximately 19 FPS. The proposed system offers a flexible, scalable, and industrially applicable solution for enhancing workplace safety in dynamic environments.

Publication Title

Frontiers in Artificial Intelligence and Applications

ISBN

[9781643686776]

Link to Full Text

Share

COinS