A Safety System for Industrial Fields using YOLO Object Detection with Deep Learning

Abstract

This paper proposes a safety system that can be used in various industrial field situations. The safety system detects boundaries with a line detection method and identifies people using YOLO (You Only Look Once) from images captured through a camera. And using the depth image, this system determines which individuals are within the danger range among the detected people. Therefore, this paper includes the selection of a specific YOLO model, performance improvement through training YOLO models with deep learning, depth data correction, line detection method, and system optimization in the proposed hardware.