Spectroscopy and Spectral Analysis, Volume. 42, Issue 5, 1572(2022)
Crop Disease Recognition Based on Visible Spectrum and Improved Attention Module
Figures & Tables(14)
Fig. 4. Crop disease recognition model based on improved channel attention mechanism
Fig. 7. Comparison results of different disease identification methods
Fig. 9. Results of heat map analysis of early and late samples of different tomato diseases
Table 1. Basic information of two Dataset
View tableView in ArticleTable 1. Basic information of two Dataset
数据集 样本图像 类别 作物样本 疾病样本 Plant Village 54 306 38 14 24 AI Challenger 2018 35 861 59 10 27
Table 2. Parameter table of SMLP_ResNet disease model
View tableView in ArticleTable 2. Parameter table of SMLP_ResNet disease model
参数 SMLP_ResNet18 SMLP_ResNet50 SMLP_ResNet101 n 2 3 3 m 2 4 4 l 2 6 23 w 2 3 3
Table 3. Recognition results of different disease models
View tableView in ArticleTable 3. Recognition results of different disease models
方法 准确率
/%训练时长/轮
/min模型权重
/M精确率
/%AlexNet[ 8 ]97.82 - 218.0 - AlexNet[ 5 ]99.08 - 218.0 - GoogleNet22[ 8 ]98.36 - - - ResNet18 99.05 4.7 42.8 98.57 SENet18 99.19 5.0 43.1 99.00 SMLP_ResNet18 99.32 4.5 48.6 99.10
Table 4. Accuracy comparison with pre-training models in previous studies
View tableView in ArticleTable 4. Accuracy comparison with pre-training models in previous studies
方法 发表年份 数据集 图像数量/张 模型 准确率/% Too et al. [ 9 ]2019 PlantVillage 54 306 VGG16 81.83 Gensheng et al.[ 10 ]2019 茶叶病害 4 980 VGG16 90.00 Wang et al.[ 11 ]2017 PlantVillage 54 306 VGG16 90.40 Agarwal M et al.[ 12 ]2020 番茄叶片病害 18 160 VGG16 93.50 Wang et al.[ 11 ]2017 PlantVillage 54 306 Inception-V3 80.00 Gandhi et al. 2018 PlantVillage 56 000 Inception-V3 88.60 Agarwal M et al.[ 12 ]2020 番茄叶片病害 18 160 Inception-V3 77.50 Elhassouny & Smarandache 2019 番茄叶片病害 7 176 Mobilenet 88.40 Gandhi et al. 2018 PlantVillage 56 000 Mobilenet 92.00 Agarwal M et al.[ 12 ]2020 番茄叶片病害 18 160 Mobilenet 82.60 Agarwal M et al.[ 12 ]2020 番茄叶片病害 18 160 CNN 98.40 Proposed Model 2021 PlantVillage 54 306 SMLP_ResNet18 99.32
Table 5. Parameter table of SMLP_ResNet disease model
View tableView in ArticleTable 5. Parameter table of SMLP_ResNet disease model
方法 准确率/% 模型权重大小/M 参数量/百万 AlexNet 83.50 217.51 57.02 ResNet18 83.83 42.75 11.21 SENet18 84.53 42.95 11.26 SMLP_ResNet18 86.93 48.58 12.73
Tools
Get Citation
Copy Citation Text
Wen-bin SUN, Rong WANG, Rong-hua GAO, Qi-feng LI, Hua-rui WU, Lu FENG. Crop Disease Recognition Based on Visible Spectrum and Improved Attention Module[J]. Spectroscopy and Spectral Analysis, 2022, 42(5): 1572
Paper Information
Category: Research Articles
Received: Mar. 15, 2021
Accepted: --
Published Online: Nov. 10, 2022
The Author Email: Wen-bin SUN (wenbinsun@mail2.gdut.edu.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20