Publication

Abstract : Accurate detection is essential for prompt reaction and fire mitigation since fires represent a serious hazard to people’s lives and property. A family of deep learning models known as Convolution Neural Network (CNN) has demonstrated excellent potential in computer vision applications, such as fire detection. The abstract illustrates how CNNs may be used to accurately detect fire in a variety of situations by learning discriminative characteristics from video frames. A comparative study has been done on Multi-Layer Perceptron (MLP) and CNN. Additionally, it examines the training procedure, the significance of data augmentation, and the structural elements of a typical CNN-based fire detection system.