返回
Improved Lightweight YOLOv11 Algorithm for Real-Time Forest Fire Detection
DOI:10.3390/electronics14081508.png)
摘要
En 中文
现代森林火灾检测的计算机视觉技术面临计算效率与检测精度之间的权衡问题。为解决此问题,我们提出了一种基于轻量级YOLOv11n(You Only Look Once version eleven)框架,该框架针对边缘部署进行了优化。骨干网络整合了一种新颖的C3k2MBNV2(Cross Stage Partial Bottleneck with 3 convolutions and kernel size 2 MobileNetV2)模块,通过紧凑架构实现高效的火灾特征提取。我们进一步在骨干网络和颈部结构中引入了SCDown(Spatial-Channel Decoupled Downsampling)模块,以在降采样过程中保留关键信息。颈部结构还集成了C3k2WTDC(Cross Stage Partial Bottleneck with 3 convolutions and kernel size 2, combined with Wavelet Transform Depthwise Convolution)模块,在减少计算开销的同时增强上下文理解。在森林火灾数据集上的实验表明,与YOLOv11n相比,我们的模型参数量减少了53.2%,浮点运算次数(FLOPs)减少28.6%,同时平均精度均值(mAP)提升3.3%。这些改进在效率与精度之间建立了最佳平衡,使所提框架能够在资源受限的边缘设备上实现实时检测能力。本研究为在低延迟和极低计算资源需求的场景中部署可靠的森林火灾检测系统提供了实用解决方案。
Keyword:
forest fire detection
YOLOv11
deep learning
lightweight
期刊
IF:
2.6
论文数:
1.0W
被引数:
4.7W
机构
引用论文
Convolutional Neural Networks Based Fire Detection in Surveillance Videos基于卷积神经网络的监控视频火灾检测
IEEE ACCESS
IF3.6
Fire and smoke real-time detection algorithm for coal mines based on improved YOLOv8s基于改进YOLOv8s的煤矿火灾与烟雾实时检测算法
PLOS ONE
IF0
FF-net: A target detection method tailored for mid-to-late stages of forest fires in complex environmentsFf-net: 一种为复杂环境下森林火灾中后期量身定制的目标检测方法
Adaptive Hierarchical Multi-Headed Convolutional Neural Network With Modified Convolutional Block Attention for Aerial Forest Fire Detection
IEEE ACCESS
IF3.6

