A fire detection algorithm based on improved YOLOv5

This paper proposes a fire detection algorithm based on improved YOLOv5 to address the issues of delayed and prone to false alarms in traditional fire systems. The proposed algorithm adopts the anchor-free object detection method, which reduces the number of parameters of the deep learning network model to improve reasoning speed. In addition, this proposed algorithm improves the precision of target detection by improving YOLOv5 backbone network structure, neck structure and detection head. The experimental simulation results show that the detection accuracy of proposed algorithm can reach 72.2%, which can effectively realize the intelligent fire detection.