International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 12 Issue: 04 | Apr 2025
p-ISSN: 2395-0072
www.irjet.net
Bridging AI and Analytics: A Smart Monitoring System for PPE Compliance in Hazardous Workspaces Riya1, Vanshita Das2, Akansha Tiwari3, Vivek Krishna Mishra4 3Vivek Kumar Mishra Professor, Dept of Computer Science and Information Technology, Dronacharya Group of
Institutions, Greater Noida, India ---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - This study introduces a YOLOv8-based PPE Compliance Detection System that uses the presence or lack of necessary personal protective equipment (PPE) to guarantee worker safety in industrial settings. We make use of a subset of the SH17 dataset, which comprises 2,350 carefully chosen photos with annotations for 6 PPE-related classifications. LabelImg was used to accomplish the annotation in YOLO format. To maximize training speed and model performance, images were scaled to 640 x 640 pixels. On Google Colab, we used a structured separation of training and validation data to train the YOLOv8n model. The Flask web application, which allows users to upload photos and obtain annotated outputs indicating PPE compliance and associated risk levels, is coupled with the trained model to give real-time inference.
human supervisors find it difficult to consistently check on each employee's compliance in large, complex work environments, particularly during high-risk activities or shift changes (Smith et al., 2021). The limitations of the traditional approaches highlight the need for an automated and intelligent solution to increase the accuracy and efficacy of PPE compliance monitoring. The accelerated development of artificial intelligence (AI) and data analytics holds an unprecedented prospect to revolutionize occupational safety processes. AI-powered smart monitoring systems utilize computer vision, machine learning, and real-time data analysis to automate detection and reporting of PPE infringements in risk-prone workplaces. Jones et al. (2022) showed that AI-powered systems are capable of detecting whether or not workers are wearing the proper protective equipment [3], including helmets, gloves, goggles, and high-visibility vests, with a high degree of accuracy. Such systems employ deep learning models trained on large databases of PPE images and actual working conditions to identify patterns and detect anomalies in real time. With the integration of predictive analytics and AI, such systems can detect non-compliance patterns, predict future safety violations, and offer meaningful recommendations to enhance workplace safety overall [4].
Key Words: PPE Detection, YOLOv8, SH17 Dataset, Object Detection, Flask Deployment, Industrial Safety, Computer Vision. 1. INTRODUCTION The foundation of occupational safety and health is ensuring adherence to personal protective equipment (PPE) in dangerous work conditions. PPE is the first line of defense against workplace dangers such as exposure to chemicals, falling objects, extreme heat, and other damaging noises. Non-adherence to PPE requirements is a significant challenge in areas like construction, manufacturing, oil and gas, and healthcare, despite the implementation of strict safety regulations and training programs. According to the International Labour Organization (ILO) [1] inadequate safety compliance and protective measures are mostly to blame for the 2.78 million occupational fatalities that are predicted to occur annually (ILO, 2023). Moreover, non-fatal diseases and injuries caused by poor compliance with PPE account for around 374 million non-fatal occupational injuries annually, causing significant economic losses and decreased productivity (ILO, 2023).
This paper discusses designing and deploying a smart monitoring system based on AI and analytics to augment PPE compliance in high-risk work environments. The envisioned system will offer real-time PPE violation detection, predictive safety insight, and automated reporting to facilitate proactive decision-making. The study aims at filling the gap between conventional safety monitoring practices and AI, proposing a scalable and effective solution towards enhancing occupational safety standards. The study also identifies the possible pitfalls of AI-based monitoring systems such as data privacy issues, biases in algorithms, and system integration issues. By overcoming these issues and tapping the potential of AI and data analytics, this study hopes to make a contribution to a safer and more compliant workplace.
Closed-circuit television (CCTV), supervisory monitoring, and manual checking are all heavily reliant on traditional PPE monitoring methods. However, these procedures are often laborious, prone to human error, small in scope, and lack real-time reaction. According to Smith et al. (2021), manual inspection systems are unreliable and unable to provide real-time feedback [2] , which leads to delayed corrective action and extended risk exposure. Furthermore,
© 2025, IRJET
|
Impact Factor value: 8.315
2. PROBLEM STATEMENT "Safety isn’t expensive, it’s priceless." As we live in a vast city where the construction industry faces several challenges that impact its efficiency and growth such as Lack of Real-
|
ISO 9001:2008 Certified Journal
|
Page 1898