Crop Disease Recognition Based on Visible Spectrum and Improved
Attention Module
SUN Wen-bin2, WANG Rong1, 3, 4, GAO Rong-hua1, 3*, LI Qi-feng1, 3, WU Hua-rui1, 3, FENG Lu1, 3
1. Beijing Research Center for Information Technology in Agriculture, Beijing 100097, China
2. College of Information Engineering, Guangdong University of Technology, Guangzhou 510006, China
3. National Engineering Research Center for Information Technology in Agriculture, Beijing 100097, China
4. College of Information Engineering, Northwest Agriculture and Forestry University, Yangling 712100, China
Abstract：Automatic identification and diagnosis of crop diseases based on visible spectrum is a challenging research field that agricultural sectors have given great attention. However, the existing research of disease identification based on convolutional neural network is prone to sacrifice network depth for the results of single disease detection, which usually resulting in the waste of hardware resources. This paper proposes a low-level crop disease identification model based on visible spectrum and improved attention mechanism, which designs a novel attention module SMLP (Squeeze-Multi-Layer-Perceptron) and crop disease identification model SMLP_ResNet. This network first uses a convolutional layer to replace the full connect layer to construct a less-parameter residual network (ResNet) whilst improving residual block (SMLP_Res) by the SMLP, Batch Normalization (BN) module and residual block (Res_block). The SMLP consists of global pooling layers and multi-layer perceptron, which can help to build a dependent relationship between channels. The multi-layer perceptron utilizes a three-layer network to double the channel dimension of global features, then restores the channel to the original dimension through carrying out a twice dimensionality reduction operation to minimize the loss of global features. On the above basis, the SMLP_Res can recalibrate the features of disease and reduce the redundant information that is useless for the detection target, and finally constructs a crop disease identification model SMLP_ResNet reduces the number of layers and enhances the accuracy of identification at the same time. Moreover, the proposed network is verified on two datasets with multiple crops and diseases, AI Challenger 2018 and Plant Village. The experimental results show that the SMLP_ResNet model achieves a high classification accuracy at the 18, 50 and 101-layer of the network, especially SMLP_ResNet18 has the best performance with accuracies up to 86.93% and 99.32% on two verification datasets respectively. SMLP_ResNet18 not only performs better than unimproved ResNet18 and SENet18 but also has higher accuracy than models in other research works with a small model size 48.6 MB, which is only 1/5 of the amount of AlexNet network of parameters, achieving a good balance between depth of network and accuracy of identification. In addition, from the heatmap conducted by Grad-CAM, it is obvious that the SMLP_ResNet18 proposed in this paper focuses more on the diseased area of leaves compared with the other networks and assigns fewer weights to the healthy area and background. In summary, the proposed model SMLP_ResNet18 realizes a high accuracy with less parameters and increases the discriminating ability to detect diseased areas and mitigates the impact of background and redundant information. Hence, this proposed model can cope with the highly precise identification of various crops.
[1] Johannes A, Picon A, Alvarez-Gila A, et al. Computers and Electronics in Agriculture, 2017, 138: 200.
[2] Liu B, Zhang Y, He D J, et al. Symmetry, 2018, 10(1): 1.
[3] Amara J, Bouaziz B, Algergawy A, et al. Gesellschaft Für Informatik, Bonn: Lecture Notes in Informatics, 2017: 79.
[4] Brahimi M, Boukhalfa K, Moussaoui A. Applied Artificial Intelligence, 2017, 31(4): 299.
[5] Sladojevic S, Arsenovic M, Anderla A, et al. Computational Intelligence and Neuroscience, 2016，2016: 1.
[6] SUN Jun, TAN Wen-jun, MAO Han-ping, et al(孙 俊，谭文军，毛罕平，等). Journal of Agricultural Engineering(农业工程学报), 2017, 33(19): 209.
[7] Hu J, Shen L, Sun G. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2020, 42(8): 2011.
[8] Mohanty S P, Hughes D P, Salathé M. Frontiers in Plant Science，2016, 7: 1.
[9] Too E C, Yujian L, Njuki S, et al. Computers and Electronics in Agriculture，2019，161：272.
[10] Hu G, Wu H, Zhang Y, et al. Computers and Electronics in Agriculture，2019，163：104852.
[11] Wang G, Sun Y, Wang J. Computational Intelligence and Neuroscience，2017，2017: 2917356.
[12] Agarwal M, Gupta S K, Biswas K K. Sustainable Computing: Informatics and Systems, 2020, 28: 100407.
[13] Selvaraju R R, Cogswell M, Das A, et al. International Journal of Computer Vision, 2020, 128(2): 336.
