Abstract: Research shows that many workers construction is one of the high-risk industries where construction workers tend to be hurt in the work process. Head injuries are very serious and often fatal. Every worker needs to wear a helmet while working in a factory or any construction site. But many workers are ignored and do work without safety equipment. The management tried to control this problem manually but it is insufficient for the real situation. The ideal solution is to develop an electronic detection system that can be automated recognize this kind of problem without human cost. The motivation for this project is to prevent the death of workers due to head injuries by monitoring real-time if a person is wearing helmet while working using Deep Learning techniques. Here, a robust approach is tried, in which CCTV cameras are used to capture the image of humans. The proposed system uses YOLO (You look only once) v3 model that is the used as the state-of-art method for real-time detection with higher rate of accuracy. The detected humans as objects are utilized for calculating the distance between them the rough Euclidean distance calculation method. The proposed model produces reliable outcome compared with the other prediction systems. We use Convolutional Neural Network (CNN) to identify who are workers are wearing helmet while entering a premise. YOLO Dark net is used to get the dependencies.

Keywords: Convolution neural network, Helmet Detection System, Image processing, real time object detection, Machine learning, YOLO, Deep Learning.

PDF | DOI: 10.17148/IJARCCE.2022.11468

Open chat
Chat with IJARCCE