Abstract: Among all kinds of disasters, fire is one of the most frequent and common major disasters that threaten public safety and social development. At present, the widely used smoke sensor method to detect fire is susceptible to factors such as distance, resulting in untimely detection. With the development of computer vision technology, image detection technology based on machine learning has been superior to traditional detection methods in terms of detection accuracy and speed, and has gradually become the emerging mainstream in the field of fire detection. At this stage, most of the methods proposed in related studies are based on high-performance hardware devices, which limits the practical application of relevant results. This paper proposes an improved fire detection algorithm based on the YOLOv5 model to address the common issues of high memory usage, slow detection speed, and high operating costs in current fire detection algorithms. The algorithm introduces FasterNet network into the backbone network to reduce model memory usage and improve detection speed. Using Ghost-Shuffle Convolution (GSConv) in the neck network reduces the number of model parameters and computational costs. Introducing a one-time aggregation cross-stage partial network module (VoV-GSCSP) to enhance feature extraction capability and improve the detection accuracy of the model. The experimental results show that compared with the original YOLOv5 model, the improved model achieves better recognition performance, with an average accuracy of 98.3%, a 31.4% reduction in memory usage, and a 13% increase in detection speed. The number of parameters decreased by 33%, and the computational workload decreased by 35%. The improved algorithm can achieve fast and accurate identification of fires, and the lightweight model is more suitable for the deployment and implementation of general embedded hardware.

Dawei Zhang and Yutang Chen. “Lightweight Fire Detection Algorithm Based on Improved YOLOv5”. International Journal of Advanced Computer Science and Applications (IJACSA) 15.6 (2024). http://dx.doi.org/10.14569/IJACSA.2024.0150681

@article{Zhang2024,
title = {Lightweight Fire Detection Algorithm Based on Improved YOLOv5},
journal = {International Journal of Advanced Computer Science and Applications},
doi = {10.14569/IJACSA.2024.0150681},
url = {http://dx.doi.org/10.14569/IJACSA.2024.0150681},
year = {2024},
publisher = {The Science and Information Organization},
volume = {15},
number = {6},
author = {Dawei Zhang and Yutang Chen}
}