Implementation of CNN-Based Computer Vision for Personal Protective Equipment Detection in the Oil and Gas Industry

Due to the inherently hazardous nature of operations in the oil and gas industry, strict compliance with safety protocols, including the obligatory use of Personal Protective Equipment (PPE), is essential for all workers. Nonetheless, monitoring PPE compliance through manual observation remains inefficient and prone to error, especially in expansive, intricate work settings. To address this challenge, there is a growing demand for an intelligent system capable of accurately and instantaneously detecting PPE use. This research introduces a Computer Vision approach employing Convolutional Neural Networks (CNNs) to identify PPE usage among workers within oil and gas environments. The system leverages a comprehensive dataset of images of workers wearing various types of PPE, including helmets, safety vests, and face masks. These images are used to train a CNN model designed to distinguish and classify the safety equipment. Experimental results demonstrate that the proposed CNN model achieves an impressive 94.2% detection accuracy on the validation data and maintains reliable performance across varying lighting conditions and camera angles. Moreover, the system can identify PPE violations in under 1 second per frame, making it suitable for real-time surveillance applications. As a result, this solution offers a promising enhancement to workplace safety oversight, with the potential to markedly reduce accident rates in the industry. The findings also pave the way for future integration with IoT-based monitoring platforms and further refinement of model adaptability across diverse industrial scenarios. The primary innovation of this study lies in the optimized deployment of CNNs tailored to the challenging conditions of oil and gas sites, delivering high detection precision and rapid response times. This area has seen limited exploration in existing literature.