Early detection of fire is critical for preventing large-scale disasters, minimizing property damage, and ensuring public safety. Traditional fire detection systems mainly rely on physical sensors such as smoke, heat, and gas detectors. Although these methods are widely used, they often suffer from delayed detection, high false alarm rates, and limitations in complex environments such as industrial facilities and crowded urban areas. With the rapid advancement of computer vision and deep learning technologies, intelligent image-based fire detection systems have emerged as an effective alternative for improving early fire detection. This paper proposes a deep learning-based intelligent fire detection and early warning system that uses Convolutional Neural Networks (CNN) to automatically identify fire in images captured from surveillance cameras. The proposed system analyses visual features from images and classifies them into two categories: fire and non-fire. A structured dataset containing fire and non-fire images is used to train and validate the deep learning model. Data augmentation techniques such as image rotation, scaling, and horizontal flipping are applied to improve model generalization and reduce overfitting. In addition, training optimization techniques including Early Stopping and ReduceLROnPlateau are implemented to enhance model performance and stability. Experimental results demonstrate that the CNN-based model significantly outperforms traditional machine learning techniques such as Logistic Regression, K-Nearest Neighbor (KNN), and AdaBoost. The proposed model achieves high classification accuracy while maintaining strong recall and AUC performance metrics. Furthermore, the system integrates an automated alarm mechanism that generates an alert when fire is detected, enabling rapid emergency response. The proposed approach provides a cost-effective, reliable, and scalable fire detection solution that can be deployed in surveillance systems for buildings, industrial environments, and smart city infrastructures. The results indicate that deep learning-based visual fire detection systems can significantly enhance disaster prevention and safety monitoring capabilities
Professor et al. (Thu,) studied this question.