An improved multiscale fusion dense network with efficient multiscale attention mechanism for apple leaf disease identification

Highlights

	● A multiscale fusion dense network with the EMA mechanism is proposed for apple leaf disease identification.
	● Replace the shallow feature extraction layer with a multiscale fusion and compare the performance of two different multiscale methods
	● Integrate the EMA mechanism into models based on the comparison among three types of attention mechanisms.
	● An improved DenseNet is proposed based on DenseNet_121, reducing half of the parameters.

Abstract

With the development of smart agriculture, accurately identifying crop diseases through visual recognition techniques instead of by eye has been a significant challenge. This study focused on apple leaf disease, which is closely related to the final yield of apples. A multiscale fusion dense network combined with an efficient multiscale attention (EMA) mechanism called Incept_EMA_DenseNet was developed to better identify eight complex apple leaf disease images. Incept_EMA_DenseNet consists of three crucial parts: the inception module, which substituted the convolution layer with multiscale fusion methods in the shallow feature extraction layer; the EMA mechanism, which is used for obtaining appropriate weights of different dense blocks; and the improved DenseNet based on DenseNet_121. Specifically, to find appropriate multiscale fusion methods, the residual module and inception module were compared to determine the performance of each technique, and Incept_EMA_DenseNet achieved an accuracy of 95.38%. Second, this work used three attention mechanisms, and the efficient multiscale attention mechanism obtained the best performance. Third, the convolution layers and bottlenecks were modified without performance degradation, reducing half of the computational load compared with the original models. Incept_EMA_DenseNet, as proposed in this paper, has an accuracy of 96.76%, being 2.93%, 3.44%, and 4.16% better than Resnet50, DenseNet_121 and GoogLeNet, respectively, proved to be reliable and beneficial, and can effectively and conveniently assist apple growers with leaf disease identification in the field.

Graphical abstract

Keywords

Incept_EMA_DenseNet / multi-scale fusion module / efficient multiscale attention mechanism / apple leaf disease identification

References

[1] Luo X W, Liao J, Zang Y, Ou Y G, Wang P. Developing from mechanized to smart agricultural production in China. Strategic Study of CAE, 2022, 24(1): 46−54 (in Chinese)

[2] Restrepo-Arias J F, Branch-Bedoya W J, Awad G . Image classification on smart agriculture platforms: systematic literature review. Artificial Intelligence in Agriculture, 2024, 13: 1–17

[3] Zhai Z, Cao Y, Xu H, Yuan P, Wang H. Review of key techniques for crop disease and pest detection. Transactions of the Chinese Society for Agricultural Machinery, 2021, 52(7): 1−18 (in Chinese)

[4] Zhao J L, Du S Z, Huang L S . Monitoring wheat powdery mildew (Blumeria graminis f. sp. tritici) using multisource and multitemporal satellite images and support vector machine classifier. Smart Agriculture, 2022, 4(1): 17–28

[5] Nettleton F D, Katsantonis D, Kalaitzidis A, Sarafijanovic-Djukic N, Puigdollers P, Confalonieri R . Predicting rice blast disease: machine learning versus process-based models. BMC Bioinformatics, 2019, 20(1): 514

[6] Raj N, Perumal S, Singla S, Sharma G K, Qamar S, Chakkaravarthy A P . Computer-aided agriculture development for crop disease detection by segmentation and classification using deep learning architectures. Computers & Electrical Engineering, 2022, 103: 108357

[7] Yu X, Yang M, Zhang H, Li D, Tang Y, Yu X. Research and application of crop diseases detection method based on transfer learning. Transactions of the Chinese Society for Agricultural Machinery, 2020, 51(10): 252−258 (in Chinese)

[8] Wang Y, Li W, Xu X, Qiu C, Wu T, Wei Q, Ma F, Han Z . Progress of apple rootstock breeding and its use. Horticultural Plant Journal, 2019, 5(5): 183–191

[9] Iftikhar M, Kandhro I A, Kausar N, Kehar A, Uddin M, Dandoush A . Plant disease management: a fine-tuned enhanced CNN approach with mobile app integration for early detection and classification. Artificial Intelligence Review, 2024, 57(7): 167

[10] Thaseentaj S, Ilango S S . Deep convolutional neural networks for south Indian mango leaf disease detection and classification. Computers, Materials & Continua, 2023, 77(3): 3593–3618

[11] Fu M, Lu C, Mao Y, Zhang X, Wu Y, Luo H, Liu Z, Li W, Ou G . An efficient and accurate deep learning method for tree species classification that integrates depthwise separable convolution and dilated convolution using hyperspectral data. International Journal of Digital Earth, 2024, 17(1): 2307999

[12] Singh A K, Rao A, Chattopadhyay P, Maurya R, Singh L . Effective plant disease diagnosis using Vision Transformer trained with leafy-generative adversarial network-generated images. Expert Systems with Applications, 2024, 254: 124387

[13] Khan M A, Akram T, Sharif M, Awais M, Javed K, Ali H, Saba T . CCDF: automatic system for segmentation and recognition of fruit crop diseases based on correlation coefficient and deep CNN features. Computers and Electronics in Agriculture, 2018, 155: 220–236

[14] Nagachandrika B, Prasath R, Joe I R P . An automatic classification framework for identifying type of plant leaf diseases using multi-scale feature fusion-based adaptive deep network. Biomedical Signal Processing and Control, 2024, 95: 106316

[15] Chen X, Shi D, Zhang H, Pérez J A S, Yang X, Li M . Early diagnosis of greenhouse cucumber downy mildew in seedling stage using chlorophyll fluorescence imaging technology. Biosystems Engineering, 2024, 242: 107–122

[16] Radočaj P, Radočaj D, Martinović G . Image-based leaf disease recognition using transfer deep learning with a novel versatile optimization module. Big Data and Cognitive Computing, 2024, 8(6): 52

[17] Liu M, Liang H, Hou M . Research on cassava disease classification using the multi-scale fusion model based on EfficientNet and attention mechanism. Frontiers in Plant Science, 2022, 13: 1088531

[18] Wen C, He W, Wu W, Liang X, Yang J, Nong H, Lan Z . Recognition of mulberry leaf diseases based on multi-scale residual network fusion SENet. PLoS One, 2024, 19(2): e0298700

[19] Zeng T, Li C, Zhang B, Wang R, Fu W, Wang J, Zhang X . Rubber leaf disease recognition based on improved deep convolutional neural networks with a cross-scale attention mechanism. Frontiers in Plant Science, 2022, 13: 829479

[20] Zhou H, Su Y, Chen J, Li J, Ma L, Liu X, Lu S, Wu Q . Maize leaf disease recognition based on improved convolutional neural network ShuffleNetV2. Plants, 2024, 13(12): 1621

[21] Wang G, Xie R, Mo L, Ye F, Yi X, Wu P . Multifactorial tomato leaf disease detection based on improved YOLOv5. Symmetry, 2024, 16(6): 723

[22] Liu W, Yu L, Luo J . A hybrid attention-enhanced DenseNet neural network model based on improved U-Net for rice leaf disease identification. Frontiers in Plant Science, 2022, 13: 922809

[23] Li D, Zhang C, Li J, Li M, Huang M, Tang Y M C C M . Multi-scale feature extraction network for disease classification and recognition of chili leaves. Frontiers in Plant Science, 2024, 15: 1367738

[24] @H. Classification of apple leaf diseases. Fei Jiang AI studio, 2022. Available at Fei Jiang AI studio website on October 20, 2024 (in Chinese)

[25] Yang Q, Shu D, and Li W . Efficient Identification of Apple Leaf Diseases in the Wild Using Convolutional Neural Networks. Agronomy, 2022, 12(11): 2784

[26] Hu J, Jiang X, Gao J, Yu X . LFMNet: a lightweight model for identifying leaf diseases of maize with high similarity. Frontiers in Plant Science, 2024, 15: 1368697

[27] Cai Z, Ou Y, Ling Y, Dong J, Lu J, Lee H . Feature detection and matching with linear adjustment and adaptive thresholding. IEEE Access: Practical Innovations, Open Solutions, 2020, 8: 189735–189746

[28] Yang L, Yu X, Zhang S, Long H, Zhang H, Xu S, Liao Y . GoogLeNet based on residual network and attention mechanism identification of rice leaf diseases. Computers and Electronics in Agriculture, 2023, 204: 107543

[29] Wang Y, Zhao G, Xiong K, Shi G . MSFF-Net: multi-scale feature fusing networks with dilated mixed convolution and cascaded parallel framework for sound event detection. Digital Signal Processing, 2022, 122: 103319

[30] Liu B, Huang X, Sun L, Wei X, Ji Z, Zhang H . MCDCNet: multi-scale constrained deformable convolution network for apple leaf disease detection. Computers and Electronics in Agriculture, 2024, 222: 109028

[31] Tian Y, Li E, Liang Z, Tan M, He X . Diagnosis of typical apple diseases: a deep learning method based on multi-scale dense classification network. Frontiers in Plant Science, 2021, 12: 698474

[32] Wang K, Jiang P, Meng J, Jiang X . Attention-based DenseNet for pneumonia classification. IRBM, 2022, 43(5): 479–485

[33] Gera D, Balasubramanian S . Landmark guidance independent spatial-channel attention and complementary context information based facial expression recognition. Pattern Recognition Letters, 2021, 145: 58–66

[34] Hao W, Ren C, Han M, Li F, Liu Z . Cattle body detection based on YOLOv5-EMA for precision livestock farming. Animals, 2023, 13(22): 3535

[35] Chen B, Zhang Z, Liu N, Tan Y, Liu X, Chen T . Spatiotemporal convolutional neural network with convolutional block attention module for micro-expression recognition. Information, 2020, 11(8): 380

[36] Zaryabi E H, Moradi L, Kalantar B, Ueda N, Halin A A . Unboxing the black box of attention mechanisms in remote sensing big data using XAI. Remote Sensing, 2022, 14(24): 6254

Acknowledgements

This work was fully supported by the National Natural Science Foundation of China (52072412).

Compliance with ethics guidelines

Dandan Dai and Hui Liu declare that they have no conflicts of interest or financial conflicts to disclose. This article does not contain any studies with human or animal subjects performed by any of the authors.

RIGHTS & PERMISSIONS

The Author(s) 2024. Published by Higher Education Press. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0)