基于改进ResNet网络的烟丝输送带洁净度分类模型
Cleanliness classification model for tobacco conveyor belt based on an improved residual network
-
摘要: 针对目前卷烟厂烟丝输送带洁净度评估依赖人工主观判断且主观性较强的问题,提出一种基于ResNet网络的改进烟丝输送带洁净度分类模型(ResNet24_SC_Block)。该模型在ResNet网络的基础上,将残差结构的网络深度设计为24层,并在残差模块中引入SE和CBAM注意力机制,以提高模型对输送带颜色及粘连烟垢等特征的捕捉能力。利用烟丝输送带数据集对模型进行验证,结果表明:ResNet24_SC_Block模型的准确率、精确率、召回率和F1分别为98.8%、98.8%、98.8%和98.4%,相较于ResNet18模型和ResNet34模型提高了3.3%~3.8%,相较于GoogLeNet和RegNet等模型提高了2.1%～17.9%,参数量比ResNet34模型减少了31.6%。该模型能够快速且准确地评估烟丝输送带洁净度,对卷烟制造厂智能化升级具有重要意义和实际应用价值。Abstract: Addressing the current reliance on manual subjective judgment and strong subjectivity for assessing the cleanliness of tobacco conveyor belts, a cleanliness classification model (ResNet24_SC_Block) for tobacco conveyor belt using an improved ResNet was proposed. This model uses ResNet for classification with a network depth of 24 layers. SE and CBAM attention mechanisms were introduced into the residual module to improve the model's ability to capture features such as conveyor belt color and adhesion smoke scale. Using the tobacco leaf conveyor belt dataset to experiment with this model, the experimental results showed that the average values of Accuracy, Precision, Recall and F1 of the improved ResNet24_SC_Block model were 98.8%, 98.8% and 98.4%, respectively, which were 3.3%～3.8% higher than those of ResNet18 model and ResNet34 model. Compared with classic and newer networks such as GoogLeNet model and RegNet model, it improves by 2.1% to 17.9%. And the number of model parameters was reduced by 31.6% compared with ResNet34 model. This approach offered notable advantages in accurately and efficiently assessing the cleanliness level of tobacco conveyor belts, making it highly consequential and practically valuable for intelligent upgrades in cigarette manufacturing plants.
-
-
[1]
吕飞,马帅旭,杨剑锋.烟草行业制造过程质量风险来源探析[J].现代企业文化,2011(3):112-113.
-
[2]
田芳,赵光辉.中国智能制造实证研究:以烟草产业为例[J].中国市场,2018(19):7-11.
-
[3]
郑贞珍.提升烟草行业智慧治理效能推进行业高质量发展的策略分析[J].经济管理,2023(7):51-54.
-
[4]
郭庆梅,于恒力,王中训,等.基于卷积神经网络的图像分类模型综述[J].电子技术应用,2023,49(9):31-38.
-
[5]
付永民,范磊,李长进,等.基于计算机视觉与机器学习的烟丝杂质图像级联检测方法[J].轻工学报,2023,38(4):113-121.
-
[6]
LU M Y,JIANG S W,WANG C,et al.Tobacco leaf grading based on deep convolutional neural networks and machine vision[J].Journal of the ASABE,2022,65(1):11-22.
-
[7]
谢裕睿,苗晟,张铄,等.基于残差神经网络的烟草病害识别研究[J].现代计算机,2020(30):27-31.
-
[8]
NIU Q F,LIU J P,JIN Y,et al.Tobacco shred varieties classification using Multi-Scale-X-ResNet network and machine vision[J].Frontiers in Plant Science,2022,13:962664.
-
[9]
李海燕,李郸,马慧宇,等.基于改进深度学习模型IRCNN的卷烟真伪鉴别[J].计算技术与自动化,2023,42(1):188-192.
-
[10]
WANG C Y,ZHAO J L,YU Z C,et al.Real-time foreign object and production status detection of tobacco cabinets based on deep learning[J].Applied Sciences,2022,12(20):10347.
-
[11]
HE K M,ZHANG X Y,REN S Q,et al.Deep residual learning for image recognition[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).Las Vegas:IEEE,2016:770-778.
-
[12]
IOFFE S,SZEGEDY C.Batch normalization:Accelerating deep network training by reducing internal covariate shift[J].JMLR,2015,37:448-456.
-
[13]
HU J,SHEN L,SUN G.Squeeze-and-excitation networks[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Salt Lake City:IEEE,2018:7132-7141.
-
[14]
WOO S,PARK J,LEE J Y,et al.CBAM:Convolutional block attention module[M].Berlin:Springer International Publishing,2018:3-19.
-
[15]
于营,杨婷婷,杨博雄.混淆矩阵分类性能评价及Python实现[J].现代计算机,2021(20):70-73,79.
-
[1]
-
点击查看大图
计量
- PDF下载量: 14
- 文章访问数: 586
- 引证文献数: 0
下载:
