Tomato Leaf Disease Classification Using EfficientNet Enhanced with RefConv Structural Reparameterization, Efficient Channel Attention, and Lion Optimizer
Abstract
Fine-grained classification of tomato leaf diseases faces challenges due to high inter-class similarity and insufficient model generalization ability. To address these issues, this paper proposes an improved EfficientNet architecture that fuses structural reparameterization, a lightweight channel attention mechanism, and an efficient optimization strategy. Specifically, the reparameterized refocus convolution (RefConv) module is introduced to enhance the extraction ability of subtle lesion textures. The original Squeeze-and-Excitation (SE) module is replaced with the efficient channel attention (ECA) mechanism to achieve lightweight channel feature recalibration. Meanwhile, the Lion optimizer is adopted to accelerate convergence and improve generalization performance on complex agricultural images. Experimental results show that the proposed method achieves 98.44% accuracy on the self-built tomato disease dataset, which is 4.22% higher than that of the baseline model EfficientNet-B0. Cross-species generalization tests further obtain 98.43% and 99.51% accuracy on apple and grape leaf disease datasets, respectively, verifying the robustness and transferability of the model.References
[1] S. P. Mohanty, D. P. Hughes, and M. Salathé (2016). Using deep learning for image-based plant disease detection, Frontiers in plant science, 7, 1419, https://doi.org/10.3389/fpls.2016.01419.
[2] J. Eunice, D. E. Popescu, M. K. Chowdary, and J. Hemanth (2022). Deep learning-based leaf disease detection in crops using images for agricultural applications, Agronomy, 12(10), 2395, https://doi.org/10.3390/agronomy12102395.
[3] O. Attallah (2023). Tomato leaf disease classification via compact convolutional neural networks with transfer learning and feature selection, Horticulturae, 9(2), 149, https://doi.org/10.3390/horticulturae9020149.
[4] M. Jelali (2024). Deep learning networks-based tomato disease and pest detection: a first review of research studies using real field datasets, Frontiers in Plant Science, 15, 1493322, https://doi.org/10.3389/fpls.2024.1493322.
[5] L. K. Ndovie and E. Masabo (2024). Leveraging MobileNetV3 for in-field tomato disease detection in Malawi via CNN, SAIEE Africa Research Journal, 115(3), 74–85, https://doi.org/10.23919/saiee.2024.10551304.
[6] M. Qabulio, M. S. Memon, S. Iqbal, P. Kumar, and A. Tsetse (2024). Effective tomato leaf disease identification model using mobilenetv3small, International Journal of Information Systems and Computer Technologies, 3(1), 57–72, https://doi.org/10.58325/ijisct.003.01.0079.
[7] R. Rashid, W. Aslam, R. Aziz, and G. Aldehim (2025). A modified mobileNetv3 coupled with inverted residual and channel attention mechanisms for detection of tomato leaf diseases, IEEE Access, https://doi.org/10.1109/ACCESS.2025.3550205.
[8] H. Sun, I. T. Nicholaus, R. Fu, and D. K. Kang (2024). YOLO-FMDI: A lightweight YOLOv8 focusing on a multi-scale feature diffusion interaction neck for tomato pest and disease detection, Electronics, 13(15), 2974, https://doi.org/10.3390/electronics13152974.
[9] W. Hou, C. Gong, W. Cao, et al. (2025). Tomato leaf disease detection method based on super-resolution enhancement and improved YOLOv8, Transactions of the Chinese Society of Agricultural Engineering, 41, 211–220, https://doi.org/10.11975/j.issn.1002-6819.202503209.
[10] U. Islam, G. Ferraioli, and V. Pascazio (2025). Tomato leaf detection, segmentation, and extraction in real-time environment for accurate disease detection, AgriEngineering, 7(4), 120, https://doi.org/10.3390/agriengineering7040120.
[11] Y. Cao (2025). Accurate Tomato Leaf Disease Identification Method based on Improved Swin Transformer, Journal of Imaging Science & Technology, 69(5), 1, https://doi.org/10.2352/J.ImagingSci.Technol.2025.69.5.050503.
[12] Y. Sun, L. Ning, B. Zhao, and J. Yan (2024). Tomato leaf disease classification by combining EfficientNetv2 and a Swin transformer, Applied Sciences, 14(7), 7472, https://doi.org/10.3390/app14177472.
[13] S. Yu, L. Xie, and L. Dai (2025). ST-CFI: Swin Transformer with convolutional feature interactions for identifying plant diseases, Scientific Reports, 15(1), 25000, https://doi.org/10.1038/s41598-025-08673-0.
[14] O. Nabil and C. Leghris (2026). EcoTomHybridNet: Policy-Guided Adaptive CNN–Transformer Inference for Resource-Aware Edge-Based Tomato Leaf Disease Classification, Future Internet, 18(5), 271, https://doi.org/10.3390/fi18050271.
[15] M. E. Qureshi, M. H. Ashraf, M. W. Arshad, A. Khan, and H. Ali (2025). Hierarchical Feature Fusion With Inception V3 for Multiclass Plant Disease Classification, Informatica, 49(27), 145–160, https://doi.org/10.31449/inf.v49i27.8208.
[16] M. Tan and Q. Le (2019). Efficientnet: Rethinking model scaling for convolutional neural networks, International Conference on Machine Learning, 97, 6105–6114.
[17] Z. Cai, X. Ding, Q. Shen, and X. Cao (2025). RefConv: Reparameterized Refocusing Convolution for Powerful ConvNets, IEEE Transactions on Neural Networks and Learning Systems, 36, 11617-11631, https://doi.org/10.1109/TNNLS.2025.3552654.
[18] Q. Wang, B. Wu, P. Zhu, et al. (2020). ECA-Net: Efficient channel attention for deep convolutional neural networks, Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 11534-11542, https://doi.org/10.1109/CVPR42600.2020.01155.
[19] X. Chen, C. Liang, D. Huang, et al. (2023). Symbolic discovery of optimization algorithms, Advances in Neural Information Processing Systems, 36, 49205-49233.
[20] C. Szegedy, W. Liu, Y. Jia, et al. (2015). Going deeper with convolutions, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 1-9.
[21] H. Ran, S. Wen, S. Wang, et al. (2020). Memristor-based edge computing of ShuffleNetV2 for image classification, IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 40(8), 1701–1710, https://doi.org/10.1109/TCAD.2020.3022970.
[22] S. H. Gao, M. M. Cheng, K. Zhao, X. Y. Zhang, M. H. Yang, and P. Torr (2019). Res2net: A new multi-scale backbone architecture, IEEE Transactions on Pattern Analysis and Machine Intelligence, 43(2), 652–662.
[23] G. Huang, Z. Liu, L. Van Der Maaten, and K. Q. Weinberger (2017). Densely connected convolutional networks, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 4700–4708.
[24] M. Sandler, A. Howard, M. Zhu, A. Zhmoginov, and L. C. Chen (2018). Mobilenetv2: Inverted residuals and linear bottlenecks, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 4510–4520.
[25] I. Radosavovic, R. P. Kosaraju, R. Girshick, K. He, and P. Dollár (2020). Designing network design spaces, Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 10428–10436.
[26] K. Han, Y. Wang, Q. Tian, J. Guo, C. Xu, and C. Xu (2020). Ghostnet: More features from cheap operations, Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 1580–1589.
DOI:
https://doi.org/10.31449/inf.v50i14.14124Keywords:
Tomato disease recognition; EfficientNet; structural reparameterization; efficient channel attention; Lion optimizer; agricultural image classificationDownloads
Published
Issue
Section
License
Authors retain copyright in their work. By submitting to and publishing with Informatica, authors grant the publisher (Slovene Society Informatika) the non-exclusive right to publish, reproduce, and distribute the article and to identify itself as the original publisher.
All articles are published under the Creative Commons Attribution license CC BY 3.0. Under this license, others may share and adapt the work for any purpose, provided appropriate credit is given and changes (if any) are indicated.
Authors may deposit and share the submitted version, accepted manuscript, and published version, provided the original publication in Informatica is properly cited.
