基于改进YOLOv5s的输电通道下的烟雾识别  

Smoke Recognition Under Transmission Channel Based on Improved YOLOv5s

作　　者：刘昶 孟琳 焦良葆[1,2] 黄国恒 吴继薇 LIU Chang;MENG Lin;JIAO Liangbao;HUANG Guoheng;WU Jiwei(AI Industrial Technology Research Institute,Nanjing Institute of Technology,Nanjing 211167,China;Jiangsu Intelligent Perception Technology and Equipment Engineering Research Center,Nanjing 211167,China)

机构地区：[1]南京工程学院人工智能产业技术研究院,南京211167 [2]江苏省智能感知技术与装备工程研究中心,南京211167

出　　处：《计算机测量与控制》2024年第12期172-177,共6页Computer Measurement &Control

基　　金：江苏省政策引导类计划项目(SZ-SQ2020007);江苏省产学研合作项目(BY20230656)。

摘　　要：针对输电通道下出现火灾险情而难以及时发现的问题,能够在火灾初期发现形状不规则且稀薄的烟雾的产生,对于险情的控制具有重要作用;为解决此问题,提出了改进YOLOv5s网络的烟雾识别算法;该方法通过在YOLOv5s模型中引入卷积注意力模块(CBAM),提高了对外轮廓并不明显的烟雾的特征提取能力;同时引入CARAFE特征上采样算法,扩大感知域,利用图片中的其他信息帮助捕捉烟雾的深层特征;为捕捉到图像中目标较小的烟雾形态,利用FReLU替换原有激活函数SiLU,用二维漏斗激活函数,在引入少量计算和过拟合风险的情况下来对网络空间中的不敏感信息进行激活,进而改善视觉任务;实验结果表明,该项目改进后的检测效果相对于原始YOLOv5s网络中的查准率提高了6.8%,查全率提高了2.8%,平均精度均值提高了2.3%,检测精度提升较为明显,更有利于应用于实际情况下的烟雾检测。In response to the challenge of detecting fire hazards in power transmission corridors in a timely manner,especially in the early stages of a fire when irregular and thin smoke is difficult to be detected,which is of great significance for controlling dangerous situations.To solve this issue,an improved smoke recognition algorithm for YOLOv5s network is proposed.A Convolutional Block Attention Module(CBAM)is introduced into the YOLOv5s model to extract the features of smoke with less distinct outlines.Additionally,the CARAFE feature upsampling algorithm is incorporated to expand the perception field and leverage other image information for capturing deep smoke features.To capture smaller smoke patterns in the images,the Sigmoid Linear Unit(SiLU)is replaced with the original activation function Funnel ReLU(FReLU),a two-dimensional funnel-shaped activation function is used to activate insensitive information in the network space while introducing minimal computational overhead and overfitting risks,thereby enhancing visual task performance.Experimental results demonstrate that the improved algorithm in this project increases the precision by 6.8%,the recall rate by 2.8%,and the mean average precision by 2.3%relative to the original YOLOv5s network.This significant enhancement in detection accuracy makes it more suitable for practical smoke detection applications.

关 键 词：输电通道 机器视觉 深度学习 注意力模块 YOLOv5s 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]