Abstract

Fire outbreaks pose significant risks to lives and property, necessitating the development of accurate and efficient detection systems. Traditional fire detection methods, such as smoke and heat sensors, often suffer from high false alarm rates and slow response times. To address these limitations, this study explores the integration of deep learning techniques with heat signature analysis to improve fire detection accuracy and reliability. The proposed system employs convolutional neural networks (CNNs) trained on a dataset comprising both infrared and visible- spectrum images, allowing for early fire detection with minimized false positives. Additionally, the study investigates the impact of environmental factors on detection performance and explores the potential of real-time monitoring using edge computing. The experimental results demonstrate that deep learning models outperform conventional fire detection techniques, offering improved precision, recall, and computational efficiency. This research provides valuable insights into the application of artificial intelligence in fire safety and highlights future directions for advancing AI-driven fire detection systems.