Leaf Disease Recognition of Horticultural Crops Based on Transfer Learning and Its Application

doi:10.11924/j.issn.1000-6850.casb2020-0430

Abstract

Abstract:

To provide farmers with recognition services for horticultural crop leaf disease more conveniently and economically, a model training method based on transfer learning and a Web deployment method based on Flask are proposed. The PlantVillage dataset is preprocessed, and three recognition models are obtained by transfer learning using ResNet18, ResNet50 and ResNet152 models. Then, these three models are deployed to the server using Flask. The average recognition accuracy of the three models for 26 leaf diseases of 14 horticultural crops, such as apple, is 95.61%, 96.63% and 97.33%, respectively, and the recognition time of a single image is 10.9, 17.9 and 33.7 ms, respectively. Considering the characteristics of the model and users’ expectation, three fast, standard and accurate recognition patterns are designed to realize the stable operation of the deep model in the server, which has some practical value.

Key words: horticultural crop, disease recognition, ResNet, transfer learning, web deployment

CLC Number:

TP391.41
S43

Li Bo, Jiang Zhaohui, Xie Jun, Rao Yuan, Zhang Wu. Leaf Disease Recognition of Horticultural Crops Based on Transfer Learning and Its Application[J]. Chinese Agricultural Science Bulletin, 2021, 37(7): 138-143.

Figures/Tables 8

类别	类别名称	类别	类别名称	类别	类别名称	类别	类别名称
0	无叶片背景	10	玉米-大斑病	20	辣椒-健康叶	30	番茄-早疫病
1	苹果-黑腐病	11	玉米-健康叶	21	马铃薯-早疫病	31	番茄-晚疫病
2	苹果-锈病	12	葡萄-黑腐病	22	马铃薯-晚疫病	32	番茄-叶霉病
3	苹果-疮痂病	13	葡萄-黑痘病	23	马铃薯-健康叶	33	番茄-斑枯病
4	苹果-健康叶	14	葡萄-叶枯病	24	树莓-健康叶	34	番茄-二斑叶螨病
5	蓝莓-健康叶	15	葡萄-健康叶	25	大豆-健康叶	35	番茄-轮斑病
6	樱桃-白粉病	16	柑橘-黄龙病	26	南瓜-白粉病	36	番茄-黄曲叶病
7	樱桃-健康叶	17	桃-细菌性斑点病	27	草莓-叶焦病	37	番茄-花叶病
8	玉米-灰斑病	18	桃-健康叶	28	草莓-健康叶	38	番茄-健康叶
9	玉米-锈病	19	辣椒-细菌性斑点病	29	番茄-细菌性斑点病

层名	输出	18层	50层	152层
conv1	112×112	7×7,64, stride 2
conv2	56×56	3×3 max pool, stride 2
conv2	56×56	$3 × 3, 3 × 3, 6464$ ×2	$1 × 1, 3 × 3, 1 × 1, 6464256$ ×3	$1 × 1, 3 × 3, 1 × 1, 6464256$ ×3
conv3	28×28	$3 × 3, 3 × 3, 128128$ ×2	$1 × 1, 3 × 3, 1 × 1, 128128512$ ×4	$1 × 1, 3 × 3, 1 × 1, 128128512$ ×8
conv4	14×14	$3 × 3, 3 × 3, 256256$ ×2	$1 × 1, 3 × 3, 1 × 1, 2562561024$ ×6	$1 × 1, 3 × 3, 1 × 1, 2562561024$ ×36
conv5	7×7	$3 × 3, 3 × 3, 512512$ ×2	$1 × 1, 3 × 3, 1 × 1, 5125122048$ ×3	$1 × 1, 3 × 3, 1 × 1, 5125122048$ ×3
1×1		Average pool, 1000-d fc, softmax
FLOPS		1.8×10⁹	3.8×10⁹	11.3×10⁹

层名	输出	18层	50层	152层
conv1	112×112	7×7,64, stride 2
conv2	56×56	3×3 max pool, stride 2
conv2	56×56	$3 × 3, 3 × 3, 6464$ ×2	$1 × 1, 3 × 3, 1 × 1, 6464256$ ×3	$1 × 1, 3 × 3, 1 × 1, 6464256$ ×3
conv3	28×28	$3 × 3, 3 × 3, 128128$ ×2	$1 × 1, 3 × 3, 1 × 1, 128128512$ ×4	$1 × 1, 3 × 3, 1 × 1, 128128512$ ×8
conv4	14×14	$3 × 3, 3 × 3, 256256$ ×2	$1 × 1, 3 × 3, 1 × 1, 2562561024$ ×6	$1 × 1, 3 × 3, 1 × 1, 2562561024$ ×36
conv5	7×7	$3 × 3, 3 × 3, 512512$ ×2	$1 × 1, 3 × 3, 1 × 1, 5125122048$ ×3	$1 × 1, 3 × 3, 1 × 1, 5125122048$ ×3
1×1		Average pool, 1000-d fc, softmax
FLOPS		1.8×10⁹	3.8×10⁹	11.3×10⁹

模型	训练集准确率	验证集准确率	测试集准确率
ResNet18	94.34	96.05	95.61
ResNet50	94.60	96.71	96.63
ResNet152	95.63	97.26	97.33

References 25

[1]	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(S1):313-317.
[2]	Hang J, Zhang D, Chen P, et al. Classification of Plant Leaf Diseases Based on Improved Convolutional Neural Network[J]. Sensors, 2019,19(19):4161.
[3]	Sladojevic S, Arsenovic M, Anderla A, et al. Deep neural networks based recognition of plant diseases by leaf image classification[J]. Computational Intelligence and Neuroscience, 2016.
[4]	孙俊, 谭文军, 毛罕平, 等. 基于改进卷积神经网络的多种植物叶片病害识别[J]. 农业工程学报, 2017,33(19):209-215.
[5]	刘永波, 雷波, 曹艳, 等. 基于深度卷积神经网络的玉米病害识别[J]. 中国农学通报, 2018,34(36):159-164.
[6]	Dechant C, Wiesnerhanks T, Chen S, et al. Automated Identification of Northern Leaf Blight-Infected Maize Plants from Field Imagery Using Deep Learning[J]. Phytopathology, 2017,107(11):1426-1432.
[7]	孙云云, 江朝晖, 董伟, 等. 基于卷积神经网络和小样本的茶树病害图像识别[J]. 江苏农业学报, 2019,35(1):48-55.
[8]	郑一力, 张露. 基于迁移学习的卷积神经网络植物叶片图像识别方法[J]. 农业机械学报, 2018,49(S1):354-359.
[9]	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.
[10]	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):230-236,253.
[11]	杨林楠, 郜鲁涛, 林尔升, 等. 基于Android系统手机的甜玉米病虫害智能诊断系统[J]. 农业工程学报, 2012,28(18):163-168.
[12]	刘洋, 冯全, 王书志. 基于轻量级CNN的植物病害识别方法及移动端应用[J]. 农业工程学报, 2019,35(17):194-204.
[13]	Zhang P, Yang L, Li D. EfficientNet-B4-Ranger: A novel method for greenhouse cucumber disease recognition under natural complex environment[J]. Computers and Electronics in Agriculture, 2020,176:105652.
[14]	Waheed A, Goyal M, Gupta D, et al. An optimized dense convolutional neural network model for disease recognition and classification in corn leaf[J]. Computers and Electronics in Agriculture, 2020,175:105456.
[15]	Krizhevsky A, Sutskever I, Hinton G E, et al. Image Net Classification with Deep Convolutional Neural Networks[A]. Advances in Neural Information Processing Systems [C]. 2012:1097-1105.
[16]	He K M, Zhang X Y, Ren S Q, et al. Deep residual learning for image recognition[A]. IEEE Conference on Computer Vision and Pattern Recognition [C]. 2016:770-778.
[17]	石祥滨, 房雪键, 张德园, 等. 基于深度学习混合模型迁移学习的图像分类[J]. 系统仿真学报, 2016,28(1):167-173,182.
[18]	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.
[19]	杨观赐, 杨静, 李少波, 等. 基于Dopout与ADAM优化器的改进CNN算法[J]. 华中科技大学学报:自然科学版, 2018,46(7):122-127.
[20]	崔义新. 基于交叉熵的随机赋权网络[D]. 保定:河北大学, 2017.
[21]	王沛. 基于Flask框架的创新创业平台系统的设计与实现[D]. 济南:山东大学, 2018.
[22]	都飞翔, 杨静, 史健芳. 基于Flask与树莓派的智能交通控制平台[J]. 现代电子技术, 2018,41(13):88-91.
[23]	Si Z, Shen K. Research on the WebP image format[A]. //Advanced Graphic Communications, Packaging Technology and Materials[M]. Springer, Singapore, 2016:271-277.
[24]	Chauhan J, Kaafar M A, Mahanti A. The Web for Underpowered Mobile Devices: Lessons Learned from Google Glass[J]. IEEE Internet Computing, 2018,22(3):38-47.
[25]	刘波, 郭申. 基于卷积定理的人脸验证CNN模型加速[J]. 北京工业大学学报, 2017,43(11):1673-1680.