Enhancing Home Safety with Automated Gas Leakage Detection Using Computer Vision

Gas leaks are a serious danger to family safety, with possible effects ranging from health risks to catastrophic tragedies. Traditional gas leak detection systems rely on manual checks of gas meter readings, which are time-consuming, susceptible to human error, and need continual monitoring by skilled staff. This study describes an automated method for detecting small gas leaks in minimum time, utilizing a state of the art deep learning model based on the YOLO (You Only Look Once) algorithm and analyzing photos of gas meter data. The suggested system takes gas meter photos, extracts digits by CNN based multi-digit recognition, and highlights crucial reading points, such as the final round bar, for further study. By comparing ongoing meter readings, the device detected abnormalities that indicate gas leakage with high accuracy. Experimentally, the system achieves mAP@50 of 99%, recall of 98%, and F1 score of 97%, which indicates high accuracy for detection of gas leaks. The improved precision analysis of the last digit enhances the sensitivity to minor leak so that it is discovered in a timely manner. The results showed that the proposed technology significantly reduces the need for human inspections while increasing the reliability of gas leak detection and providing a reliable option for continuous monitoring and home safety.