基于CA-MobileNet-V2的核桃病害识别与应用

doi:10.3969/j.issn.1004-1524.20221763

浙江农业学报 ›› 2023, Vol. 35 ›› Issue (12): 2977-2987.DOI: 10.3969/j.issn.1004-1524.20221763

基于CA-MobileNet-V2的核桃病害识别与应用

李荣鹏(), 买买提·沙吾提(), 盛艳芳, 何旭刚

新疆大学地理与遥感科学学院,新疆乌鲁木齐 830017;b. 新疆大学新疆绿洲生态重点实验室,新疆乌鲁木齐 830017;c. 新疆大学智慧城市与环境建模自治区普通高校重点实验室,新疆乌鲁木齐 830017

收稿日期:2022-12-12 出版日期:2023-12-25 发布日期:2023-12-27
作者简介:李荣鹏(1996—),男,山西晋中人,硕士研究生,主要从事农业遥感图像分类识别研究。E-mail:lrpwyyxx@163.com
通讯作者: *买买提·沙吾提,E-mail:korxat@xju.edu.cn
基金资助:
新疆自然科学计划(自然科学基金)联合基金项目(2021D01C055)

Identification and application of walnut disease based on CA-MobileNet-V2

LI Rongpeng(), MAMAT Sawut(), SHENG Yanfang, HE Xugang

College of Geography and Remote Sensing Sciences; b. Xinjiang Key Laboratory of Oasis Ecology; c. Key Laboratory of Smart City and Environment Modelling of Higher Education Institute, Xinjiang University, Urumqi 830017, China

Received:2022-12-12 Online:2023-12-25 Published:2023-12-27

摘要/Abstract

摘要：

病害侵袭是制约核桃优质发展的重要因素之一,为实现田间智能化病害识别,设计了一款核桃病害识别模型。该模型采用Mobilenet-V2作为基础网络骨架,在倒残差结构中添加坐标注意力机制,解决特征提取时位置信息缺失的问题。此外,设计混合迁移的训练方式,将跨域迁移和域内迁移相结合,避免单独迁移学习的不良影响。结果表明:1)混合迁移对模型提升效果最佳,准确率最高提升18.57百分点。2)模型平均识别准确率为96.97%,模型参数量为3.95 M,内存占有量为10.50 MB,相较于Mobilenet-V3-Large、ShuffulNet-V2和EfficientNet-V2-S,识别准确率分别提升4.39百分点、6.63百分点和4.31百分点,且保持较少的参数量与内存占有量。3)与SE(squeeze-and-excitation)模块、CBAM(convolutional block attention module)模块相比,坐标注意力机制更能提升模型对感兴趣区域的关注度。因此,该模型可用于开发安卓应用程序并部署于移动端,为核桃病害智能识别提供新方法。

关键词: 核桃病害, 坐标注意力机制, 混合迁移, 安卓应用程序

Abstract:

Disease invasion is one of the important factors restricting the high-quality development of walnut. In order to realize intelligent disease identification in the field, this study designed a walnut disease identification model. The model used Mobilenet-V2 as the basic network skeleton, and added a coordinate attention mechanism to the inverted residual structure to make up for the lack of location information during feature extraction. In addition, this study designed a mixed transfer training method, which combined cross-domain and intra-domain to avoid the adverse effects of separate transfer learning. The results showed that: 1) The mixed transfer had the best effect on improving the model, and the highest accuracy rate was increased by 18.57 percentage points. 2) The average identification accuracy of the model was 96.97%, the model parameter size was 3.95 M, and the memory occupancy was 10.50 MB. Compared with Mobilenet-V3-Large, ShuffulNet-V2 and EfficientNet-V2-S, the identification accuracy was increased by 4.39, 6.63 and 4.31 percentage points, respectively, and the parameter size and memory occupation were keep small. 3) Compared with SE (squeeze-and-excitation) and CBAM (convolutional block attention module), the coordinate attention mechanism could improve the model’s attention to the region of interest. Therefore, the model could be used to develop an Android application and deploy it on the mobile terminal, and provide a new method for intelligent identification of walnut disease.

Key words: walnut disease, coordinate attention mechanism, mixed transfer, Android application

中图分类号:

S436.6

李荣鹏, 买买提·沙吾提, 盛艳芳, 何旭刚. 基于CA-MobileNet-V2的核桃病害识别与应用[J]. 浙江农业学报, 2023, 35(12): 2977-2987.

LI Rongpeng, MAMAT Sawut, SHENG Yanfang, HE Xugang. Identification and application of walnut disease based on CA-MobileNet-V2[J]. Acta Agriculturae Zhejiangensis, 2023, 35(12): 2977-2987.

图/表 9

参考文献 21

[1]	奚声珂, 郗荣庭, 马杰. 世界核桃生产与研究动态[J]. 经济林研究, 1990, 8(1): 76-79.
	XI S K, XI R T, MA J. Trends of walnut production and research in the world[J]. Non-Wood Forest Research, 1990, 8(1): 76-79. (in Chinese)
[2]	王彦翔, 张艳, 杨成娅, 等. 基于深度学习的农作物病害图像识别技术进展[J]. 浙江农业学报, 2019, 31(4): 669-676.
	WANG Y X, ZHANG Y, YANG C Y, et al. Advances in new nondestructive detection and identification techniques of crop diseases based on deep learning[J]. Acta Agriculturae Zhejiangensis, 2019, 31(4): 669-676. (in Chinese with English abstract)
[3]	张建华, 孔繁涛, 吴建寨, 等. 基于改进VGG卷积神经网络的棉花病害识别模型[J]. 中国农业大学学报, 2018, 23(11): 161-171.
	ZHANG J H, KONG F T, WU J Z, et al. Cotton disease identification model based on improved VGG convolution neural network[J]. Journal of China Agricultural University, 2018, 23(11): 161-171. (in Chinese with English abstract)
[4]	FERENTINOS K P. Deep learning models for plant disease detection and diagnosis[J]. Computers and Electronics in Agriculture, 2018, 145: 311-318.
[5]	TOO E C, LI Y J, NJUKI S, et al. A comparative study of fine-tuning deep learning models for plant disease identification[J]. Computers and Electronics in Agriculture, 2019, 161: 272-279.
[6]	葛道辉, 李洪升, 张亮, 等. 轻量级神经网络架构综述[J]. 软件学报, 2020, 31(9): 2627-2653.
	GE D H, LI H S, ZHANG L, et al. Survey of lightweight neural network[J]. Journal of Software, 2020, 31(9): 2627-2653. (in Chinese with English abstract)
[7]	ELFATIMI E, ERYIGIT R, ELFATIMI L. Beans leaf diseases classification using MobileNet models[J]. IEEE Access, 2022, 10: 9471-9482.
[8]	CHEN J D, ZHANG D F, SUZAUDDOLA M, et al. Identifying crop diseases using attention embedded MobileNet-V2 model[J]. Applied Soft Computing, 2021, 113: 107901.
[9]	LU L, LIU W, YANG W B, et al. Lightweight corn seed disease identification method based on improved ShuffleNetV2[J]. Agriculture, 2022, 12(11): 1929.
[10]	孙道宗, 刘锦源, 丁郑, 等. 基于改进EfficientNetv2模型的多品种南药叶片分类方法[J]. 华中农业大学学报, 2023, 42(1): 258-267.
	SUN D Z, LIU J Y, DING Z, et al. Classification of leaves of multi-variety southern traditional Chinese medicine based on improved EfficientNetv2 model[J]. Journal of Huazhong Agricultural University, 2023, 42(1): 258-267. (in Chinese with English abstract)
[11]	SINGH P, VERMA A, ALEX J S R. Disease and pest infection detection in coconut tree through deep learning techniques[J]. Computers and Electronics in Agriculture, 2021, 182: 105986.
[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.
	LIU Y, FENG Q, WANG S Z. Plant disease identification method based on lightweight CNN and mobile application[J]. Transactions of the Chinese Society of Agricultural Engineering, 2019, 35(17): 194-204. (in Chinese with English abstract)
[13]	SANDLER M, HOWARD A, ZHU M L, et al. MobileNetV2: inverted residuals and linear bottlenecks[C]// 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. June 18-23, 2018, Salt Lake City. IEEE, 2018: 4510-4520.
[14]	NIU Z Y, ZHONG G Q, YU H. A review on the attention mechanism of deep learning[J]. Neurocomputing, 2021, 452: 48-62.
[15]	HU J, SHEN L, SUN G. Squeeze-and-excitation networks[C]// 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. June 18-23, 2018, Salt Lake City. IEEE, 2018: 7132-7141.
[16]	WOO S, PARK J, LEE J Y, et al. CBAM: convolutional block attention module[M]//Computer Vision- ECCV 2018. Cham: Springer International Publishing, 2018: 3-19.
[17]	HOU Q B, ZHOU D Q, FENG J S. Coordinate attention for efficient mobile network design[C]// 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). June 20-25, 2021, Nashville. IEEE, 2021: 13708-13717.
[18]	LIU H X, LV H C, LI J J, et al. Research on maize disease identification methods in complex environments based on cascade networks and two-stage transfer learning[J]. Scientific Reports, 2022, 12: 18914.
[19]	ZHAO X, LI K Y, LI Y X, et al. Identification method of vegetable diseases based on transfer learning and attention mechanism[J]. Computers and Electronics in Agriculture, 2022, 193: 106703.
[20]	CHEN J D, CHEN J X, ZHANG D F, et al. Using deep transfer learning for image-based plant disease identification[J]. Computers and Electronics in Agriculture, 2020, 173: 105393.
[21]	SELVARAJU R R, COGSWELL M, DAS A, et al. Grad-CAM: visual explanations from deep networks via gradient-based localization[C]// 2017 IEEE International Conference on Computer Vision (ICCV). October 22-29, 2017, Venice. IEEE, 2017: 618-626.

编辑推荐

Metrics

阅读次数

全文

490

HTML			PDF

最新录用	在线预览	正式出版	最新录用	在线预览	正式出版
0	0	10	0	0	480

来源	本网站	其他网站

次数	338	152
比例	69%	31%

摘要

181

最新录用	在线预览	正式出版

0	0	181

	来源	本网站

	次数	181
	比例	100%

病害类型 Disease category	标签 Labels	样本数量Number of samples
		增强前Before augmentation		增强后After augmentation
		数据集A Dataset A	数据集B Dataset B	数据集A Dataset A	数据集B Dataset B
虫蛀叶Insect pest	0	152	261	679	1 139
健康叶Healthy	1	123	145	616	1 012
褐斑病Brown spot	2	105	125	630	1 000
黑斑病Black spot	3	98	189	686	1 134
缺素病Element deficiency	4	162	242	656	1 210
炭疽病Anthrax	5	116	168	696	1 008
总计Total		756	1 130	3 963	6 503

病害类型 Disease category	标签 Labels	样本数量Number of samples
		增强前Before augmentation		增强后After augmentation
		数据集A Dataset A	数据集B Dataset B	数据集A Dataset A	数据集B Dataset B
虫蛀叶Insect pest	0	152	261	679	1 139
健康叶Healthy	1	123	145	616	1 012
褐斑病Brown spot	2	105	125	630	1 000
黑斑病Black spot	3	98	189	686	1 134
缺素病Element deficiency	4	162	242	656	1 210
炭疽病Anthrax	5	116	168	696	1 008
总计Total		756	1 130	3 963	6 503

迁移方式 Transfer methods	数据集 Dataset	准确率 Accuracy/ %
无迁移 No transfer	数据集B Dataset B	86.62
learning
跨域迁移 Cross-domain	ImageNet数据集+数据集B Datase ImageNet+Dataset B	93.53
transfer
域内迁移 Intra-domain	数据集A+数据集B Dataset A+Dataset B	78.40
transfer
混合迁移 Mixed transfer	ImageNet数据集+数据集A+数据集B Dataset ImageNet+Dataset A+Dataset B	96.97

迁移方式 Transfer methods	数据集 Dataset	准确率 Accuracy/ %
无迁移 No transfer	数据集B Dataset B	86.62
learning
跨域迁移 Cross-domain	ImageNet数据集+数据集B Datase ImageNet+Dataset B	93.53
transfer
域内迁移 Intra-domain	数据集A+数据集B Dataset A+Dataset B	78.40
transfer
混合迁移 Mixed transfer	ImageNet数据集+数据集A+数据集B Dataset ImageNet+Dataset A+Dataset B	96.97

模型 Models	类别 Classification	精确率 Precision/%	召回率 Recall/%	F1值 F1score/%	平均准确率 Mean accuracy%	参数量 Params/M	内存占有 Memory/MB
VGG16	虫害Insect pest	87.27	63.44	73.47	64.48	138.23	512.27
	健康Healthy	79.12	35.64	49.15
	褐斑Brown spot	51.30	49.50	50.38
	黑斑Black spot	49.54	94.25	64.94
	缺素Element deficiency	78.73	71.90	75.16
	炭疽Anthrax	68.18	67.16	67.66
ResNet18	虫害Insect pest	95.63	96.48	96.05	97.31	11.82	42.72
	健康Healthy	99.50	100.00	99.75
	褐斑Brown spot	94.17	96.04	95.10
	黑斑Black spot	99.06	94.20	96.57
	缺素Element deficiency	98.00	99.49	98.74
	炭疽Anthrax	98.02	98.51	98.26
Mobilenet-V3-Large	虫害Insect pest	87.34	92.17	89.69	92.58	5.42	21.65
	健康Healthy	96.55	96.55	96.55
	褐斑Brown spot	86.41	89.48	87.92
	黑斑Black spot	92.96	85.71	89.19
	缺素Element deficiency	93.00	93.00	93.00
	炭疽Anthrax	95.05	94.12	94.58
ShuffulNet-V2	虫害Insect pest	83.84	91.87	87.67	90.34	2.28	9.03
	健康Healthy	92.12	92.57	92.35
	褐斑Brown spot	85.92	85.55	85.71
	黑斑Black spot	92.96	85.71	89.19
	缺素Element deficiency	93.00	93.00	93.00
	炭疽Anthrax	95.05	94.12	97.57
EfficientNet-V2-S	虫害Insect pest	87.77	93.50	90.54	92.66	23.25	81.28
	健康Healthy	95.07	94.61	94.84
	褐斑Brown spot	88.35	90.55	89.44
	黑斑Black spot	94.67	88.55	91.36
	缺素Element deficiency	95.00	93.60	94.29
	炭疽Anthrax	96.04	95.57	95.80
CA-Mobilenet-V2	虫害Insect pest	95.20	96.04	95.61	96.97	3.95	10.50
	健康Healthy	99.02	99.51	99.26
	褐斑Brown spot	93.20	96.00	94.58
	黑斑Black spot	98.59	92.92	95.67
	缺素Element deficiency	98.00	99.49	98.74
	炭疽Anthrax	98.02	98.51	98.26

基于CA-MobileNet-V2的核桃病害识别与应用

Identification and application of walnut disease based on CA-MobileNet-V2

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 9

参考文献 21

相关文章 15

编辑推荐

Metrics

本文评价

模型 Model	准确率 Accuracy/ %	漏检率 Missing/ %	误检率 Error/%	耗时 Time consuming/s
CA-MobileNet-V2	90.83	4.17	5.00	0.53
ResNet18	92.50	4.17	3.33	3.37
MobileNet-V3-Large	87.50	5.83	6.67	1.05
ShuffulNet-V2	82.50	5.00	12.50	0.36
EfficientNet-V2-S	86.67	7.50	5.83	5.09

[1]	余帅, 黄俊, 应俊杰, 张娟, 毛雪琴, 张治军, 吕要斌, 李龑. 香蕉皮粉末对黑腹果蝇的引诱活性及挥发物化学成分分析[J]. 浙江农业学报, 2023, 35(10): 2415-2424.
[2]	宋传生, 康晓飞, 樊庆忠, 王俊刚, 石雪, 张子汝, 谭青青, 曾小娇, 刘芳, 李英赛, 侯常跃. 枣疯病植原体胸苷激酶基因的克隆、序列分析与原核表达[J]. 浙江农业学报, 2023, 35(8): 1763-1772.
[3]	曾雅婷, 熊桃, 李红叶. 柑橘黑点病菌(Diaporthe citri)快速分子检测技术[J]. 浙江农业学报, 2022, 34(7): 1457-1465.
[4]	杨秀娟, 李卫雅, 李彩苗, 成碧君, 高芬, 赵军. 茶籽饼对黄芪和三七根腐病病原菌的抑制效果[J]. 浙江农业学报, 2022, 34(6): 1227-1235.
[5]	吴嘉维, 姚张良, 胡琪琪, 张杰, 陈轶, 蒋建荣, 周国鑫, 王霞. 浙北桐乡梨锈病防治适期和防治药剂研究[J]. 浙江农业学报, 2021, 33(9): 1668-1675.
[6]	孟幼青, 汪恩国, 陈吴健, 李艳敏, 程帆, 孟敏霞. 浙江省亚洲柑橘木虱黄龙病病原分布与消长规律[J]. 浙江农业学报, 2021, 33(3): 464-469.
[7]	黄振东, 蒲占湑, 胡秀荣, 陈国庆, 吕佳, 占红木. 矿物油乳剂对柑橘木虱定殖行为的影响[J]. 浙江农业学报, 2021, 33(1): 87-95.
[8]	赖家豪, 宋水林, 刘冰. 三株柑橘溃疡病生防内生细菌对脐橙感染溃疡病后几种防御酶活性的影响[J]. 浙江农业学报, 2020, 32(11): 1994-2000.
[9]	张小彦, 何静, 侯彩霞, 张树衡. 枸杞根腐病菌拮抗菌株的筛选与鉴定[J]. 浙江农业学报, 2020, 32(5): 858-865.
[10]	郭雪松, 田丽波, 商桑, 邹凯茜, 陈虹容, 李婉豫, 岳晓琦. 芒果蒂腐病拮抗放线菌A10和A17的分离、鉴定和特征化研究[J]. 浙江农业学报, 2020, 32(3): 460-468.
[11]	邱晓红, 郑路敏, 褚姝频, 朱友理. 桃园3种蛀果害虫雄成虫动态及其与气温和降雨量相关性[J]. 浙江农业学报, 2020, 32(3): 469-474.
[12]	林瑞, 任海英, 安笑笑, 郑锡良, 梁森苗, 张淑文, 戚行江. 生物有机肥对杨梅凋萎病防控及其树势恢复的影响[J]. 浙江农业学报, 2019, 31(7): 1096-1104.
[13]	乔中全, 王晓明, 李永欣, 曾慧杰, 蔡能, 刘思思, 陈艺. 38个紫薇品种亲缘关系的ISSR分析[J]. 浙江农业学报, 2019, 31(4): 565-571.
[14]	孟幼青, 翁海勇, 岑海燕, 李红叶, 何勇. 潜伏期柑橘黄龙病宿主糖代谢及近红外光谱特征[J]. 浙江农业学报, 2019, 31(3): 428-435.
[15]	康海婷, 凌丹燕, 吴志娟, 路梅, 宗宇, 郭卫东. 蓝莓枝干溃疡病病原鉴定及品种抗病性研究[J]. 浙江农业学报, 2019, 31(3): 436-443.