YOLO Algorithm for Helmet Detection in Industries for Safety Purpose
N. K. Anushkannan, Vijaya Kumbhar, Suresh Kumar Maddila, Chandra Sekhar Kolli, B Vidhya, R.G. Vidhya
Abstract
Using a Safety helmet is crucial for ensuring the safety of employees in the construction, power, and manufacturing sector. However, employees frequently remove their helmets due to a lack of security awareness and inconvenience, exposing themselves to further hazards. Employees without helmets would experience higher injuries in incidents like falling, electrical shock, and dropping heavy loads. As a result, fast and accurate safety helmet detection is an absolute necessity for all management. Conventional human monitoring is time-consuming and difficult to supervise employees in a large population. Because of its low accuracy and poor resilience, the industry’s conventional helmet identification method is not recommended for use. Therefore, this research offers a deep learning approach to identifying safety helmet wearing with good accuracy. Collecting the requisite images is done using the Kaggle website. A total of 5000 images are obtained, and 1000 from that will be utilized to test the DL model. Region-based Convolutional Neural Network (RCNN) and You Only Look Once (YOLO) is selected as DL model for helmet detection. The measures such as mean Average Precision (mAP), recall, and precision are used to Figure out the best model out of two. The YOLO has achieved the greatest mAP success rate of 97.12%.