Enhancing Fire Detection Comparison using YOLOv5, Mobile Net SSD, and Custom CNN

This study tackles the intrinsic difficulties that come with fighting fire and addresses the urgent need to improve fire detection techniques. This study thoroughly assesses well-known pre-trained models, YOLOv5 and MobileNet SSD, in addition to a custom designed Convolutional Neural Network (CNN) architecture for fire detection, with an emphasis on sophisticated object detection methods. The results show that the custom CNN model performs better than MobileNet SSD (48%) and YOLOv5 (55%), with an outstanding mean Average Precision (mAP) of 70.9%. The unmatched potential of the method in enhancing fire detection applications is highlighted by this data-driven analysis.