부산대학교 도서관

The real-time fire detection and triggering system presented in this study harnesses camera images for swift and reliable fire detection, leveraging a multi-stage approach. The system’s preprocessing phase involves image enhancement and noise reduction to optimize image quality, while the feature extraction step extracts relevant information from the images to improve discrimination between fire and non-fire regions. Classification is carried out using a machine learning model, like Convolutional Neural Networks (CNNs), to discern potential fire events in the images. Upon detecting a fire, the system triggers immediate alarms or safety protocols, thus enhancing fire prevention and response mechanisms in real-time scenarios. The overall implementation of the research work is done in the python environment from which it is proved that the proposed methodology attains better performance than the existing research methodology. The proposed work attains better performance than the existing research methodologies in terms of 7% increased accuracy. And in terms of precision, recall and f-measure proposed method attains 6%, 4% improvement than the existing research methodologies.