Research and Implementation of an Agricultural Pest and Disease Detection Algorithm Based on Deep Learning

Xuesong Sha

doi:10.54097/3w2q1263

Authors

Xuesong Sha

DOI:

https://doi.org/10.54097/3w2q1263

Keywords:

Deep learning, Agricultural pest and disease detection, YOLOv11, RSC-YOLOv11-EMA, Small-object detection, Precision agriculture

Abstract

Food security is increasingly threatened by crop pests and diseases, while conventional field inspection remains labor intensive, experience dependent, and prone to missed detection during early disease stages. To support accurate and real-time pest and disease monitoring in precision agriculture, this study proposes RSC-YOLOv11-EMA, an improved lightweight detection framework for soybean leaf pest and disease images. First, YOLOv11 is selected as the baseline after comparison with YOLOv5 and YOLOv8 on the same dataset. Second, the backbone network is reconstructed with RepVGG so that multi-branch training can be converted into a single-path inference structure, reducing computational cost for edge deployment. Third, SENet channel attention is embedded into multi-scale feature outputs to strengthen discriminative lesion channels and suppress background interference. Fourth, K-means++ is used to analyze the true scale distribution of lesions and guide scale-aware augmentation for small targets. On this basis, an Efficient Multi-scale Attention module, warm-up cosine annealing, and exponential moving average weight updating are introduced to improve spatial sensitivity, convergence stability, and generalization on a small-sample dataset. Experiments show that the final RSC-YOLOv11-EMA model achieves 93.6% precision, 90.4% recall, 91.2% mAP@0.5, and 31.2 FPS, improving mAP@0.5 by 6.5 percentage points over the YOLOv11 baseline while preserving real-time inference. A PyQt5-based detection system is further implemented, with an end-to-end response time of 145 ms on edge equipment. The results demonstrate that the proposed method provides an effective and deployable solution for early agricultural pest and disease detection.

Downloads

Download data is not yet available.

References

[1] Ngugi, L. C., Abelwahab, M., & Abo-Zahhad, M. (2021). Recent advances in image processing techniques for automated leaf pest and disease recognition: A review. Information Processing in Agriculture, 8(1), 27-51. https://doi.org/10.1016/j.inpa.2020.04.003

[2] Kamilaris, A., & Prenafeta-Boldú, F. X. (2018). Deep learning in agriculture: A survey. Computers and Electronics in Agriculture, 147, 70-90.

https://doi.org/10.1016/j.compag.2018.02.016

[3] Redmon, J., Divvala, S., Girshick, R., & Farhadi, A. (2016). You only look once: Unified, real-time object detection. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 779-788). IEEE.

https://doi.org/10.1109/CVPR.2016.91

[4] Fuentes, A., Yoon, S., Kim, S. C., & Park, D. S. (2017). A robust deep-learning-based detector for real-time tomato plant diseases and pests recognition. Sensors, 17(9), 2022. https://doi.org/10.3390/s17092022

[5] Ding, X., Zhang, X., Ma, N., Han, J., Ding, G., & Sun, J. (2021). RepVGG: Making VGG-style ConvNets great again. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 13733-13742). IEEE.

https://doi.org/10.1109/CVPR46437.2021.01352

[6] Woo, S., Park, J., Lee, J. Y., & Kweon, I. S. (2018). CBAM: Convolutional block attention module. In Proceedings of the European Conference on Computer Vision (pp. 3-19). Springer. https://doi.org/10.1007/978-3-030-01234-2_1

[7] Wang, Q., Wu, B., Zhu, P., Li, P., Zuo, W., & Hu, Q. (2020). ECA-Net: Efficient channel attention for deep convolutional neural networks. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 11534-11542). IEEE.

https://doi.org/10.1109/CVPR42600.2020.01155

[8] Ouyang, D., He, S., Zhang, G., Luo, M., Guo, H., Zhan, J., & Huang, Z. (2023). Efficient multi-scale attention module with cross-spatial learning. In Proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing (pp. 1-5). IEEE.

https://doi.org/10.1109/ICASSP49357.2023.10096512

[9] Loshchilov, I., & Hutter, F. (2016). SGDR: Stochastic gradient descent with warm restarts. arXiv, arXiv:1608.03983. https://doi.org/10.48550/arXiv.1608.03983

[10] Izmailov, P., Podoprikhin, D., Garipov, T., Vetrov, D., & Wilson, A. G. (2018). Averaging weights leads to wider optima and better generalization. arXiv, arXiv:1803.05407. https://doi.org/10.48550/arXiv.1803.05407

[11] Tang, K., Qian, Y., Dong, H., Zhang, Y., Liu, S., & Wang, Z. (2025). SP-YOLO: A real-time and efficient multi-scale model for pest detection in sugar beet fields. Insects, 16(1), 102. https://doi.org/10.3390/insects16010102

[12] Xue, R., & Wang, L. (2025). Research on lightweight citrus leaf pest and disease detection based on PEW-YOLO. Processes, 13(5), 1365. https://doi.org/10.3390/pr13051365

[13] Sun, J., Feng, Z., Han, J., Liu, Y., Wang, X., & Li, H. (2025). YOLO-PLNet: A lightweight real-time detection model for peanut leaf diseases based on edge deployment. Frontiers in Plant Science, 16, 1707501.

https://doi.org/10.3389/fpls.2025.1707501

[14] Yang, X., He, Q., Xie, X., Liu, Y., & Zhang, W. (2025). YOLO-RP: A lightweight and efficient detection method for small rice pests in complex field environments. Symmetry, 17(10), 1598. https://doi.org/10.3390/sym17101598

[15] Xu, D., Xiong, H., Liao, Y., Zhao, Z., Liu, T., & Chen, J. (2024). EMA-YOLO: A novel target-detection algorithm for immature yellow peach based on YOLOv8. Sensors, 24(12), 3783. https://doi.org/10.3390/s24123783