In forest and grassland fire scenarios, the diversity of open flame forms and the complexity of the environment may lead to false or missed detection. Therefore, an improved YOLOv8n fire detection algorithm (YOLOv8n-CSA) is proposed for forest and grassland fires. CSA (channel-spatial attention) is the channel spatial attention module, and a group shuffle convolution (GSConv) module is introduced to replace the third layer standard convolution module (Conv) in the original YOLOv8n, reducing model computation and improving feature extraction ability. And introducing the Slim Neck structure in the head further reduces the computational complexity of the model. Simultaneously design a channel spatial attention module (CSA) integrated into the Backbone section to enhance the expressive power of the input feature map.This module combines channel attention, channel shuffle, and spatial attention mechanisms to capture global dependencies within feature maps. Based on a forest and grassland fire dataset, and without utilizing pretrained models, the proposed fire detection network achieves a 3.7% increase in precision, a 1.51% improvement in recall, a 3.24% enhancement in mAP50, and a 5.62% reduction in GFLOPs compared to the baseline YOLOv8n model. Experimental results demonstrate that the proposed algorithm not only reduces computational cost but also enhances the detection performance of fire-related features.
CAOL L, LIUX Y, CHENX,et al.Prediction model of forest fire occurrence probability in Yanbian area,Jilin Province[J].Journal of Northeast Forestry University,2024,52(3):90-96.
ZHANGL C, SUNZ C, DONGL B.Evaluation of recognition and restoration effect of burned areas based on Landsat time series data[J].Forest Engineering,2024,40(5):8-16.
ZHOUQ, ZHANGH, ZHANGQ L,et al.Identification and prediction models of driving factors for forest fires in Daxing'an Mountains of Inner Mongolia,northern China[J].Journal of Beijing Forestry University,2024,46(12): 114-125.
HEN L, ZHANGJ S, LINW S.Forest fire image recognition based on deep learning multi-target detection technology[J].Journal of Nanjing Forestry University(Natural Sciences Edition),2024,48(3): 207-218.
ZHANGX W, ZHANGF Q.Open flame detection method in forest scenes based on SBP-YOLOv7[J/OL].Journal of Nanjing Forestry University (Natural Science Edition),1-7[2022-12-24].
[20]
MAS, LIW, WANL,et al.A lightweight fire detection algorithm based on the improved YOLOv8 model[J].Applied Sciences,2024,14(16):6878.
CUIK B, GENGJ C.A multi scene fire sign detection algorithm based on EE-YOLOv 8s[J/OL].Journal of Graphics,1-15 [2022-12-24].
[23]
HUANGJ, ZHOUJ, YANGH,et al.A small-target forest fire smoke detection model based on deformable transformer for end-to-end object detection[J].Forests,2023,14(1):162.
[24]
LYU C, ZHOUH, CHENY,et al.A lightweight fire detection algorithm for small targets based on YOLOv5s[J].Scientific Reports,2024,14(1):14104.
[25]
CAOL, SHENZ, XUS.Efficient forest fire detection based on an improved YOLO model[J].Visual Intelligence,2024,2(1):20.
[26]
LUANT, ZHOUS, ZHANGG,et al.Enhanced lightweight YOLOX for small object wildfire detection in UAV imagery[J].Sensors,2024,24(9):2710.
[27]
WANGC Y, BOCHKOVSKIYA, LIAOH Y M.YOLOv7:Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C]//2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR),17-24 June 2023,Vancouver,BC,Canada.IEEE,2023:7464-7475.
[28]
ZHUX, CHENGD, ZHANGZ,et al.An empirical study of spatial attention mechanisms in deep networks[C]// 2019 IEEE/CVF International Conference on Computer Vision (ICCV),27 October 2019 - 02 November 2019,Seoul,Korea (South).IEEE,2019:6688-6697.
[29]
ZHANGX, ZHOUX, LINM,et al.Shufflenet:An extremely efficient convolutional neural network for mobile devices[C]// 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition,18-23 June 2018,Salt Lake City,UT,USA.IEEE,2018:6848-6856.
[30]
HUJ, SHENL, SUNG.Squeeze-and-excitation networks[C]// 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition,18-23 June 2018,Salt Lake City,UT,USA.IEEE,2018:7132-7141.
[31]
HANK, WANGY, TIANQ,et al.Ghostnet:More features from cheap operations[C]// 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR),13-19 June 2020,Seattle,WA,USA.IEEE,2020:1580-1589.
[32]
LIH, LIJ, WEIH,et al.Slim-neck by GSConv:A better design paradigm of detector architectures for autonomous vehicles[J].arXiv preprint arXiv:2022.
[33]
SHAMSOSHOARA, AFGHAHF, RAZIA,et al.Aerial imagery pile burn detection using deep learning:The FLAME dataset[J].Computer Networks,2021,193:108001.
[34]
DOLLARP, WOJEKC, SCHIELEB,et al.Pedestrian detection:a benchmark[C]//2009 IEEE Conference on Computer Vision and Pattern Recognition,20-25 June 2009,Miami FL,USA.IEEE,2009:304-311.