[1]汤文亮,黄梓锋.基于知识蒸馏的轻量级番茄叶部病害识别模型[J].江苏农业学报,2021,(03):570-578.[doi:doi:10.3969/j.issn.1000-4440.2021.03.004]
　TANG Wen-liang,HUANG Zi-feng.Lightweight model of tomato leaf diseases identification based on knowledge distillation[J].,2021,(03):570-578.[doi:doi:10.3969/j.issn.1000-4440.2021.03.004]
点击复制

基于知识蒸馏的轻量级番茄叶部病害识别模型()
分享到：

江苏农业学报[ISSN:1006-6977/CN:61-1281/TN]

卷:
期数:
2021年03期
页码:
570-578
栏目:
植物保护
出版日期:
2021-06-30

文章信息/Info

Title:
Lightweight model of tomato leaf diseases identification based on knowledge distillation
作者:
汤文亮; 黄梓锋
(华东交通大学信息工程学院,江西南昌330013)
Author(s):
TANG Wen-liang; HUANG Zi-feng
(School of Information Engineering, East China Jiaotong University, Nanchang 330013, China)
关键词:
番茄病害; 识别模型; 条件卷积; 注意力机制; 知识蒸馏
Keywords:
tomato disease; identification model; conditional convolution; attention mechanism; knowledge distillation
分类号:
TP391.41
DOI:
doi:10.3969/j.issn.1000-4440.2021.03.004
文献标志码:
A
摘要:
目前，基于迁移学习诊断农作物病害已经成为一种趋势，然而大多数研究使用的模型参数众多，占用了大量设备空间并且推理演算耗时较长，导致对存储和计算资源有严格限制的设备无法利用深度神经网络的优势。为此，本研究以PlantVillage数据集中的番茄病害样本为研究对象，基于条件卷积及通道注意力机制，提出1种新颖的轻量级模型，同时使用知识蒸馏法指导模型训练，在保证模型性能的前提下压缩模型大小。将AlexNet、VGG16、GoogLeNet、ResNet50及DenseNet121进行对比，并利用类激活图（CAM）可视化模型分类决策的图像区域。结果表明，经过蒸馏的自定义模型可以精准定位番茄病叶的发病区域，在测试集中的平均识别准确率达97.6%，不仅优于其他模型，而且模型大小仅为4.4 M。
Abstract:
At present, crop disease diagnosis based on transfer learning has become a trend. However, models used in most studies had a large number of parameters, which occupied a lot of equipment space and took much time for inference. The above conditions make devices with strict restrictions on storage and computing resources cannot take advantage of deep neural networks. Thus, tomato disease samples from the PlantVillage dataset were used as the research object in this study, a novel lightweight model based on conditional convolution and channel attention mechanism was proposed. At the same time, knowledge distillation method was used to train the custom model, which could greatly reduce the model volume while ensuring the performance of the model. The AlexNet, VGG16, GoogLeNet, ResNet50 and DenseNet121 were compared, and the class activation map (CAM) was used to visualize the image area in the model classification decision. The results showed that the distilled user-defined model could accurately locate the diseased tomato leaves. The average recognition accuracy in the test set was 97.6%, which was higher than other models and the model size was only 4.4 M.

参考文献/References:

[1]吴琼,梁巧兰,张娜. 番茄主要病害病原菌培养条件及室内药剂筛选[J]. 甘肃农业大学学报, 2018, 53(5):79-86.
[2]谭海文,吴永琼,秦莉,等. 我国番茄侵染性病害种类变迁及其发生概况[J]. 中国蔬菜, 2019, 359(1):86-90.
[3]贾少鹏,高红菊,杭潇. 基于深度学习的农作物病虫害图像识别技术研究进展[J].农业机械学报, 2019，50(增刊1):313-317.
[4]李颀,王康,强华,等. 基于颜色和纹理特征的异常玉米种穗分类识别方法[J].江苏农业学报, 2020,36(1):24-31.
[5]张经纬,贡亮,黄亦翔,等. 基于随机森林算法的黄瓜种子腔图像分割方法[J].农机化研究, 2017,39(10): 163-168.
[6]GINNE M, PUNITHA S C. Tomato disease segmentation using K-means clustering[J]. International Journal of Computer Applications, 2016, 144(5):25-29.
[7]GUO Y, LIU Y, OERLEMANS A, et al. Deep learning for visual understanding: a review[J]. Neurocomputing, 2016, 187(26):27-48.
[8]LECUN Y, BENGIO Y, HINTON G. Deep learning[J]. Nature, 2015, 521(7553):436.
[9]周飞燕,金林鹏,董军. 卷积神经网络研究综述[J]. 计算机学报, 2017,40(6):1229-1251.
[10]赵立新,侯发东,吕正超,等. 基于迁移学习的棉花叶部病虫害图像识别[J]. 农业工程学报, 2020, 36(7):184-191.
[11]黄双萍,孙超,齐龙,等. 基于深度卷积神经网络的水稻穗瘟病检测方法[J]. 农业工程学报, 2017, 33(20):169-176.
[12]许景辉,邵明烨,王一琛,等. 基于迁移学习的卷积神经网络玉米病害图像识别[J]. 农业机械学报, 2020, 51(2): 25.
[13]龙满生,欧阳春娟,刘欢,等. 基于卷积神经网络与迁移学习的油茶病害图像识别[J]. 农业工程学报, 2018, 34(18):194-201.
[14]PRABHAKAR M, PURUSHOTHAMAN R, AWASTHI D P. Deep learning based assessment of disease severity for early blight in tomato crop[J]. Multimedia Tools and Applications, 2020,79: 1-12.
[15]MOHANTY S P, HUGHES D P, SALATH M. Using deep learning for image-based plant disease detection[J]. Frontiers in Plant Science, 2016, 7:1419.
[16]郭小清,范涛杰,舒欣. 基于改进Multi-Scale AlexNet的番茄叶部病害图像识别[J]. 农业工程学报, 2019(13):162-169.
[17]YANG B, BENDER G, LE Q V, et al. Condconv: conditionally parameterized convolutions for efficient inference[C]//SCHLKOPF B, PLATT J, HOFMANN T. Advances in Neural Information Processing Systems. Vancouver, Canada：2019：1307-1318.
[18]HU J, SHEN L, ALBANIE S, et al. Squeeze-and-excitation networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 42: 2011-202399.
[19]HINTON G, VINYALS O, DEAN J. Distilling the knowledge in a neural network[J]. Computer Science, 2015, 14(7):38-39.
[20]SELVARAJU R R, COGSWELL M, DAS A, et al. Grad-CAM: Visual explanations from deep networks via gradient-based localization[J]. International Journal of Computer Vision, 2019, 128(8): 336-359.
[21]HUGHES D, SALATH M. An open access repository of images on plant health to enable the development of mobile disease diagnostics[EB/OL].（2015-11-25）
[2020-09-01]. https: //arxiv.org/abs/1511. 08060.
[22]SZEGEDY C, LIU W, JIA Y, et al. Going deeper with convolutions[C]//IEEE. Proceedings of the IEEE conference on computer vision and pattern recognition. Boston, MA:IEEE, 2015: 1-9.
[23]XIE S, GIRSHICK R, DOLLR P, et al. Aggregated residual transformations for deep neural networks[C]//IEEE. Proceedings of the IEEE conference on computer vision and pattern recognition. New York:IEEE Press, 2017: 1492-1500.
[24]RUSSAKOVSKY O, DENG J, SU H, et al. Imagenet large scale visual recognition challenge[J]. International Journal of Computer Vision, 2015, 115(3): 211-252.
[25]KRIZHEVSKY A, SUTSKEVER I, HINTON G. ImageNet classification with deep convolutional neural networks[J]. Communications of the ACM, 2017, 60(6):84-90.
[26]SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for large-scale image recognition [EB/OL].（2014-09-04）
[2020-09-01].https://arxiv.org/abs/1712.031497.
[27]HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition[C]//IEEE. Proceedings of the IEEE conference on computer vision and pattern recognition. New York: IEEE Press, 2016: 770-778.
[28]HUANG G, LIU Z, VAN DER MAATEN L, et al. Densely connected convolutional networks[C]//IEEE. Proceedings of the IEEE conference on computer vision and pattern recognition. New York:IEEE Press, 2017: 4700-4708.
[29]李懿超,沈润平,黄安奇. 基于深度学习的湘赣鄂地区植被变化及其影响因子关系模型[J]. 江苏农业科学,2019,47(3):213-218.

备注/Memo

备注/Memo:
收稿日期：2020-09-10基金项目：江西省重点研发计划重点项目（20192ACB50027）；江西省教育厅科技重点项目（GJJ190296）；江西省交通厅项目（2019X0016）作者简介：汤文亮（1969-），男，江西南昌人，硕士，教授，硕士生导师，主要从事大数据分析与深度学习研究。（E-mail）1402855900@qq.com通讯作者：黄梓锋，（E-mail）fshzfxqeng@163.com
更新日期/Last Update: 2021-07-05