Smart Gas Leak Detection using Thermal Vision and Deep Neural Networks

This study presents a web-based gas leak detection system that uses thermal imaging and deep learning techniques.
A Convolutional Neural Network (CNN) model is developed to analyze thermal images and classify them into
four categories: No Gas, Perfume, Smoke, and Mixture. The system is implemented using Python and Flask,
allowing users to upload thermal images through a secure interface and receive predictions along with confidence
scores. User data and prediction history are stored using SQLite for future reference. The proposed model
demonstrates strong performance in detecting gas leakage conditions and provides a practical solution for
improving safety in industrial and environmental settings. The system is designed to be simple, scalable, and easy
to use.