AUTHOR=Zheng Hongtao , Dembélé Sounkalo , Wu Yongxin , Liu Yan , Chen Hongli , Zhang Qiujie 

TITLE=A lightweight algorithm capable of accurately identifying forest fires from UAV remote sensing imagery

JOURNAL=Frontiers in Forests and Global Change

VOLUME=Volume 6 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/forests-and-global-change/articles/10.3389/ffgc.2023.1134942

DOI=10.3389/ffgc.2023.1134942

ISSN=2624-893X

ABSTRACT=Nowadays, deep learning algorithms are widely used in forest fire monitoring systems. In high-altitude mon-itoring, the sizes of flames are too small, and they are potentially heavily obscured by trees, making it impos-sible for the algorithm to extract more information. However, if the accuracy is improved by increasing the complexity of the algorithm, the speed of the algorithm is greatly reduced. To achieve a breakthrough in both algorithm speed and accuracy, this paper proposes a two-stage recognition method that combines the novel YOLO algorithm (FireYOLO) with Real-ESRGAN. Firstly, as regards the structure of the FireYOLO algo-rithm, “the backbone part adopts GhostNet and introduces a dynamic convolutional structure, which im-proves the information extraction capability of the morphologically variable flame while greatly reducing the computational effort; the neck part introduces a novel cross-layer connected, two-branch FPN structure, which greatly improves the information extraction capability of small targets and re-duces the loss in the in-formation transmission process; the head embeds the attention-guided module (ESNet) proposed in this pa-per, which enhances the attention capability of small targets”. Secondly, the flame region recognized by FireYOLO is input into Real-ESRGAN after a series of cropping and stitching operations to enhance the clarity, and then the enhanced image is recognized for the second time with FireYOLO, and, finally, the recognition result is overwritten back into the original image. Our experiments show that the algorithms in this paper run very well on both PC-based and embedded devices, adapting very well to situations where they are obscured by trees as well as changes in lighting. The overall recognition speed of Jeston Xavier NX is about 20.67 FPS (latency-free real-time inference), which is 21.09% higher than the AP of YOLOv5x, and is the best performance fire detection algorithm with excellent application prospects.