International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 12 Issue: 10 | Oct 2025
p-ISSN: 2395-0072
www.irjet.net
Powered Smart Monitoring For Industrial Accident Detection Using Convolutional Neural Network Akileshwari A1, Alageshwaran P2, Abishek G P3, Sanjeevi K S ⁴ 1Mrs.A.Akileshwari Associate Professor2, Dept. of Computer Science Engineering, KLN college of Engineering,
Tamil Nadu, INDIA
2Alageshwaran P, Dept. of Computer Science Engineering, KLN college of Engineering, Tamil Nadu, INDIA
3 Abishek G P, Dept. of Computer Science Engineering, KLN college of Engineering, Tamil Nadu, INDIA
⁴Sanjeevi K S, Dept. of Computer Science Engineering, KLN college of Engineering, Tamil Nadu, INDIA ---------------------------------------------------------------------***--------------------------------------------------------------------Abstract - Industrial environments are inherently 1.INTRODUCTION hazardous, with a high risk of accidents that can cause injuries, equipment damage, and operational downtime. Traditional accident reporting and monitoring methods are often manual, limited in coverage, and prone to delays, highlighting the need for intelligent real-time monitoring systems. This report presents a powered smart monitoring system for automated detection of industrial accidents using Convolutional Neural Networks (CNNs). The system combines vision-based sensors, such as CCTV or industrial cameras, with an edge computing unit running an optimized CNN model to continuously analyze live video streams. Video preprocessing techniques, including noise reduction, illumination normalization, and motion detection, are applied to improve the accuracy of frame selection before analysis by the CNN, which has been trained on a diverse set of annotated industrial accident scenarios, such as slips, falls, collisions, and machinery malfunctions. When an accident is detected with confidence above a predefined threshold, real-time alerts are instantly dispatched to supervisors and safety control systems, facilitating rapid emergency response and minimizing the impact of incidents. Experimental results demonstrate that the proposed system achieves high accuracy with low false positive rates and detects accidents with minimal latency, making it feasible for deployment in complex industrial environments. The study also addresses challenges such as varying lighting conditions, occlusions, multiple accident types, and the balance between detection speed and computational cost. Furthermore, recommendations are provided for enhancing system performance, including multi-camera data fusion, augmentation of the training dataset with rare accident cases, and hardware acceleration using GPUs or TPUs. By enabling continuous, automated, and real-time accident detection, the proposed system significantly improves industrial safety, reduces human dependency, and provides a scalable solution adaptable to various industrial settings, demonstrating the potential of deep learning-based monitoring systems to transform accident prevention strategies in hazardous workplaces.
© 2025, IRJET
|
Impact Factor value: 8.315
In modern industrial environments, ensuring worker safety and preventing accidents are of utmost importance. Traditional monitoring systems often rely on manual supervision and conventional sensors, which can be slow or ineffective in detecting hazardous situations. With the rapid advancement of Artificial Intelligence (AI) and the Internet of Things (IoT), smart monitoring systems have become a powerful alternative. This project, Powered Smart Monitoring for Industrial Accident Detection Using Convolutional Neural Network (CNN), aims to automatically detect industrial accidents such as fires, falls, or unsafe behaviors in real time using video surveillance data. By leveraging deep learning techniques, particularly CNNs, the system can analyze visual input, identify abnormal events, and alert authorities instantly. This intelligent approach not only enhances workplace safety but also minimizes response time, reduces human error, and contributes to a safer and more efficient industrial ecosystem.
1.2.OBJECTIVE AND SCOPE OF THE PROJECT The main objective of this project is to develop an intelligent monitoring system that can automatically detect industrial accidents using Convolutional Neural Networks (CNN). The system aims to ensure worker safety by continuously analyzing live video feeds from industrial areas to identify abnormal events such as fire, falls, or unsafe behaviors. By integrating deep learning with image processing, the system can recognize hazardous situations in real time and immediately send alerts to concerned authorities for quick action. The scope of this project includes designing and training a CNN model, implementing it for real-time monitoring, and creating an alert mechanism to enhance safety and reduce human dependency. This project can be further expanded with IoT-based sensors, cloud connectivity, and mobile notifications, making it a scalable and efficient solution for improving industrial safety management.
|
ISO 9001:2008 Certified Journal
|
Page 482