基于DCP-ShuffleNetV2的轻量级森林害虫识别方法

doi:10.13733/j.jcam.issn.2095-5553.2025.01.029

摘要/Abstract

摘要： 针对现有害虫识别模型复杂度高、计算量和参数量巨大的问题，提出一种基于DCP-ShuffleNetV2的轻量级森林害虫识别模型。该模型主要从特征提取、特征融合、轻量化方面进行改进。首先通过引入金字塔分割注意力模块PSA提取多尺度的空间信息和跨通道依赖关系，有效地学习上下文信息；其次将基准网络模型ShuffleNetV2的Stage模块修改为CSP结构，增强特征融合能力；将模型的普通卷积替换为动态卷积，压缩模型参数量和计算量。试验以雄安新区“千年秀林”害虫为研究对象，构建30类常见害虫数据集。结果表明，改进后的DCP-ShuffleNetV2模型在自制的Forest30数据集上的害虫识别准确率是92.43%，模型参数量、计算量和内存大小分别是0.13 M、24.53 M和9.53 MB，相比于基准网络模型，识别准确率提升3.11%，参数量、计算量和内存大小分别减少62.83%、42.48%和15.13%。与目前常用的分类模型相比，识别准确率平均提高5.39%，模型参数量、计算量和内存大小平均减小14.32 M、1 035.80 M和35.98 MB。

关键词: 害虫识别, DCP-ShuffleNetV2, 注意力机制, CSP结构, 特征提取

Abstract: This paper proposes a pest recognition model based on lightweight DCP-ShuffleNetV2 to solve the problems of high complexity of model, large amount of computation and reference. The model is improved from feature extraction, feature fusion and lightweight. Firstly, to learn context information effectively, the Pyramid Split Attention （PSA） module is introduced to extract multi-scale spatial information and cross-channel dependency. Secondly, to enhance the feature fusion capability, the Stage module of the benchmark network model ShuffleNetV2 is modified to Cross Stage Partial （CSP） structure. Finally, to compress the number of parameters and computation, the regular convolution is replaced by dynamic convolution for the model. In the experiment, a data set of Forest 30 was constructed in the “Millennium Xiulin” of Xiong'an New Area. The experimental results show that the pest identification accuracy of the DCP-ShufflenetV2 model is 92.43%, and the number of parameters, computation amount and memory size of the improved model are 0.13 M, 24.53 M and 9.53 MB, respectively. Compared with the ShufflenetV2 network model, the pest identification accuracy of the improved model increased by 3.11%, and the reference number, computation amount and memory size were reduced by 62.83%, 42.48% and 15.13%, respectively. Compared with the current commonly used classification model, the average recognition accuracy is increased by 5.39%, the number of parameters, computation amount and memory size of the improved model are reduced by 14.32 M, 1 035.80 M and 35.98 MB on average.

Key words: pest identification, DCP-ShuffleNetV2, attention mechanism, CSP structure, feature extraction

中图分类号:

S763
TP391.4

高天赐, 王克俭, 陈晨, 韩宪忠, 王超, 李会平, . 基于DCP-ShuffleNetV2的轻量级森林害虫识别方法[J]. 中国农机化学报, 2025, 46(1): 190-197.

Gao Tianci, , Wang Kejian, , Chen Chen, , Han Xianzhong, , Wang Chao, , Li Huiping, . Identification method of lightweight forest pest based on DCP-ShuffleNetV2[J]. Journal of Chinese Agricultural Mechanization, 2025, 46(1): 190-197.

参考文献

［1］　国家统计局. 2021中国统计年鉴［M］. 北京: 中国统计出版社, 2021.
［2］　翟肇裕, 曹益飞, 徐焕良, 等. 农作物病虫害识别关键技术研究综述［J］. 农业机械学报, 2021, 52（7）: 1-18.
Zhai Zhaoyu, Cao Yifei, Xu Huanliang, et al. Review of key techniques for crop disease and pest detection ［J］. Transactions of the Chinese Society for Agricultural Machinery, 2021, 52（7）: 1-18.
［3］　李文勇, 李明, 陈梅香,等. 基于机器视觉的作物多姿态害虫特征提取与分类方法［J］. 农业工程学报, 2014, 30（14）: 154-162.
Li Wenyong, Li Ming, Chen Meixiang, et al. Feature extraction and classification method of multipose pests using machine vision ［J］. Transactions of the Chinese Society of Agricultural Engineering, 2014, 30（14）: 154-162.
［4］　王美华, 吴振鑫, 周祖光. 基于注意力改进CBAM的农作物病虫害细粒度识别研究［J］. 农业机械学报, 2021, 52（4）: 239-247.
Wang Meihua, Wu Zhenxin, Zhou Zuguang. Finegrained identification research of crop pests and diseases based on improved CBAM via attention ［J］. Transactions of the Chinese Society of Agricultural Machinery, 2021, 52（4）: 239-247.
［5］　张永玲, 姜梦洲, 俞佩仕, 等. 基于多特征融合和稀疏表示的农业害虫图像识别方法［J］. 中国农业科学, 2018, 51（11）: 2084-2093.Zhang Yongling, Jiang Mengzhou, Yu Peishi, et al. Agricultural pest identification based on multifeature fusion and sparse representation ［J］. Scientia Agricultura Sinica, 2018, 51（11）: 2084-2093.
［6］　肖志云, 刘洪. 小波域马铃薯典型虫害图像特征选择与识别［J］. 农业机械学报, 2017, 48（9）: 24-31.
Xiao Zhiyun, Liu Hong. Features selection and recognition of potato typical insect pest images in wavelet domain ［J］. Transactions of the Chinese Society for Agricultural Machinery, 2017, 48（9）: 24-31.
［7］　邹修国, 丁为民, 刘德营,等. 基于4种不变矩和BP神经网络的稻飞虱分类［J］. 农业工程学报, 2013, 29（18）: 171-178.
Zou Xiuguo, Ding Weimin, Liu Deying,et al. Classification of rice planthopper based on invariant moments and BP neural network ［J］. Transactions of the Chinese Society of Agricultural Engineering, 2013, 29（18）: 171-178.
［8］　陈娟, 陈良勇, 王生生,等. 基于改进残差网络的园林害虫图像识别［J］. 农业机械学报, 2019, 50（5）: 187-195.
Chen Juan, Chen Liangyong, Wang Shengsheng,et al. Pest image recognition of garden based on improved residual network ［J］. Transactions of the Chinese Society for Agricultural Machinery, 2019, 50（5）: 187-195.
［9］　Deng L, Wang Y, Han Z, et al. Research on insect pest image detection and recognition based on bioinspired methods ［J］. Biosystems Engineering, 2018, 169: 139-148.
［10］　孔建磊, 金学波, 陶治,等. 基于多流高斯概率融合网络的病虫害细粒度识别［J］. 农业工程学报, 2020, 36（13）: 148-157.
Kong Jianlei, Jin Xuebo, Tao Zhi,et al. Finegrained recognition of diseases and pests based on multistream Gaussian probability fusion network ［J］. Transactions of the Chinese Society of Agricultural Engineering, 2020, 36（13）: 148-157.
［11］　Wei D, Chen J, Luo T, et al.Classification of crop pests based on multiscale feature fusion ［J］. Computers and Electronics in Agriculture, 2022, 194: 106736.
［12］　Khanramaki M, AsliArdeh E A, Kozegar E. Citrus pests classification using an ensemble of deep learning models ［J］. Computers and Electronics in Agriculture, 2021, 186: 106192.
［13］　Yang Z, Yang X, Li M, et al. Automated gardeninsect recognition using improved lightweight convolution network ［J］. Information Processing in Agriculture, 2021.
［14］　李江昀, 赵义凯, 薛卓尔,等. 深度神经网络模型压缩综述［J］. 工程科学学报, 2019, 41（10）: 1229-1239.〖JP2〗Li Jiangyun, Zhao Yikai, Xue Zhuoer, et al. A survey of model compression for deep neural networks ［J］. Chinese Journal of Engineering, 2019, 41（10）: 1229-1239.
［15］　Thenmozhi K, Reddy U S. Crop pest classification based on deep convolutional neural network and transfer learning ［J］. Computers and Electronics in Agriculture, 2019, 164: 104906.
［16］　Liu Y, Liu S, Xu J, et al. Forest pest identification based on a new dataset and convolutional neural network model with enhancement strategy ［J］. Computers and Electronics in Agriculture, 2022, 192: 106625.
［17］　Kim Y, Koh Y J, Lee C, et al. Dark image enhancement based on pairwise target contrast and multiscale detail boosting ［C］. International Conference on Image Processing （ICIP）. IEEE, 2015: 1404-1408.
［18］　Ma N, Zhang X, Zheng H T, et al. ShuffleNetV2: Practical guidelines for efficient CNN architecture design ［C］. Proceedings of the European Conference on Computer Vision， 2018: 116-131.
［19］　Chen Y, Dai X, Liu M, et al. Dynamic convolution: Attention over convolution kernels ［C］. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020: 11030-11039.
［20］　Zhang H, Zu K, Lu J, et al. EPSANet: An efficient pyramid squeeze attention block on convolutional neural network ［J］. arXiv preprint arXiv:210514447, 2021.
［21］　Sandler M, Howard A, Zhu M, et al. MobileNetV2: Inverted residuals and linear bottlenecks ［C］. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018: 4510-4520.
［22］　Wang C Y, Liao H Y M, Wu Y H, et al. CSPNet: A new backbone that can enhance learning capability of CNN ［C］. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2020: 390-391.
［23］　Wang Q, Wu B, Zhu P, et al. ECANet: Efficient channel attention for deep convolutional neural networks ［C］. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2020: 11531-11539.
［24］　Hou Q, Zhou D, Feng J. Coordinate attention for efficient mobile network design ［C］. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021: 13713-13722.
［25］　Hu J, Shen L, Sun G. Squeezeandexcitation networks ［C］.Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018: 7132-7141.
［26］　Selvaraju R R, Cogswell M, Das A, et al. GradCAM: Visual explanations from deep networks via gradientbased localization ［C］. Proceedings of the IEEE International Conference on Computer Vision, 2017: 618-626.

[1]	高芳征, 温鑫, 黄家才, 陈光明, 金少宇, 赵雪迪. 基于AD-YOLOX-Nano的茶叶嫩芽识别算法[J]. 中国农机化学报, 2025, 46(1): 178-184.
[2]	王鑫淼, 张正, 董晓威, 王林烽, 李瑞祥. 基于改进YOLOv8算法的谷子田杂草检测[J]. 中国农机化学报, 2025, 46(1): 185-189.
[3]	卢明, 余心杰, 郭俊先. 基于改进YOLOv8算法的水稻田间人工施肥行为识别[J]. 中国农机化学报, 2025, 46(1): 220-226.
[4]	韩鹏飞, 宋其江, 贾梦实. 基于改进轻量化 EfficientNet-V2模型的小麦种子分类[J]. 中国农机化学报, 2024, 45(9): 111-117.
[5]	张传栋, 亓璐, 丁华立. 基于改进 YOLOv8n模型的多品种葡萄簇检测方法[J]. 中国农机化学报, 2024, 45(9): 220-226.
[6]	李洁, 高尚兵, 余骥远, 陈新, 李士丛, 袁星星. 基于PMMS-Net和叶绿素荧光成像的绿豆叶斑病抗性鉴定方法[J]. 中国农机化学报, 2024, 45(8): 210-216.
[7]	刘星, 顾寄南, 黄则栋, 张文浩, 张伟. 基于深度学习的苹果点云语义分割方法研究[J]. 中国农机化学报, 2024, 45(8): 223-227.
[8]	苏斐, 王光辉, 史艳霞, 贾然, 闫银发, 祖林禄, . 融合通道注意力机制的ResUnet作物病斑分割模型[J]. 中国农机化学报, 2024, 45(8): 228-233.
[9]	冯晓, 张辉, 刘运超, 张微, 李小红, 马中杰, . 基于改进YOLOv5s的硬核期葡萄簇检测[J]. 中国农机化学报, 2024, 45(8): 240-245.
[10]	周思捷, 刘天奇, 陈天华. 基于改进YOLOv5算法的水稻病害识别研究[J]. 中国农机化学报, 2024, 45(8): 246-253.
[11]	王克涛, 陈世锋, 陈贵, 韦锦, 蒙丽雯, 陈泉成. 基于YOLOv5s的园林球形绿篱检测方法[J]. 中国农机化学报, 2024, 45(8): 262-269.
[12]	王雷, 钟康生, 郭小宝, 盛斌, 肖波, 荆磊. 基于机器视觉的陈皮热泵干燥含水率预测模型研究[J]. 中国农机化学报, 2024, 45(7): 97-103.
[13]	岳耀华, 张伟, 亓立强, . 基于改进YOLOv5的田间大豆花朵生长状态检测方法研究[J]. 中国农机化学报, 2024, 45(7): 188-193.
[14]	吴利刚, 吕媛媛, 周倩, 陈乐, 张梁, 史建华. 基于改进YOLOv5的轻量级黄花成熟检测方法[J]. 中国农机化学报, 2024, 45(7): 235-242.
[15]	雷浩, 苑迎春, 何振学. 基于多尺度卷积和注意力机制的枣品种识别[J]. 中国农机化学报, 2024, 45(6): 135-141.