Transforming Plant Disease Diagnosis: Utilising TPUs-accelerated CNNs and Visual Transformers for Quick and Accurate Detection of Soybean Leaf Diseases
Keywords:
Soybean plants, PyTorch XLA, TPU accelerationsAbstract
Our work addresses the critical requirement for accurate and quick plant disease diagnosis, especially in agricultural contexts where prompt detection may significantly reduce crop losses. we researched how well Tensor Processing Units (TPUs)-accelerated Convolutional Neural Networks (CNNs) perform in providing a quick and precise diagnosis of illnesses affecting soybean leaves. Enhancing crop management techniques and increase food security. Drawing on recent advances with regard to Visual Transformers (VTs) as noted by Wu et al., we suggest a new approach that uses transformers instead of traditional convolutions to connect semantic ideas in token-space. Integration of the PyTorch XLA library with supervised algorithm for TPU acceleration, we create a reliable CNN model specifically designed for accurate leaf disease diagnosis. After extensive testing and 80 epochs of cross-validation, our model yields an outstanding accuracy of 88.37% with a low total loss of 0.2315. Moreover, we accelerate the training process by using TPUs, improving the effectiveness and accessibility of research. This study highlights the transformational power in the field of plant pathology. The visual transformer employed has a fundamental patch size of 16x16 and an image resolution of 384x384 pixels.
Downloads
References
Wu, B., Xia, S., Huang, W., & Zhang, C. (2020). Visual Transformers: Token-based Image Representation and Processing for Computer Vision. arXiv. Retrieved from https://arxiv.org/abs/2006.03677
Chowdhury, M. E. H., Rahman, T., Khandakar, A., Ayari, M. A., Khan, A. U., Khan, M. S., Al-Emadi, N., Reaz, M. B. I., Islam, M. T., & Ali, S. H. M. (2021). Automatic and reliable leaf disease detection using deep learning techniques. AgriEngineering, 3(2), 294–312. https://doi.org/10.3390/agriengineering3020020
You, Y., Zhang, Z., Hsieh, C.-J., Demmel, J., & Keutzer, K. (2019). Fast deep neural network training on distributed systems and cloud TPUs. IEEE Transactions on Parallel and Distributed Systems, 30(11), 2449–2462. https://doi.org/10.1109/TPDS.2019.-2913833
Ahad, M. T., Li, Y., Song, B., & Bhuiyan, T. (2023). Comparison of CNN-based deep learning architectures for rice diseases classification. Artificial Intelligence in Agriculture, 9, 22–35. https://doi.org/10.1016/j.aiia.2023.07.001
Trippa, D., Scalenghe, R., Basso, M. F., Panno, S., Davino, S., Morone, C., Giovino, A., Oufensou, S., Luchi, N., Yousefi, S., & Martinelli, F. (2023). Next-generation methods for early disease detection in crops. Pest Management Science, 00(00), 1–10. https://doi.org/10.1002/ps.7733
Gao, Y., & Liu, C. (2022). Study on classification system of rape root tumors based on TPU. In 2022 IEEE Asia-Pacific Conference on Image Processing, Electronics, and Computers (IPEC) (pp. 1049–1055). IEEE. https://doi.org/10.1109/IPEC54454.2022.-9777609
Liu, Y., Zhou, Z., Han, K., Tian, Q., & Wang, Y. (2024). A survey of visual transformers. IEEE Transactions on Neural Networks and Learning Systems, 35(6), 7478–7498. https://doi.org/10.1109/TNNLS.2022.3227717
Yu, J., Zhang, J., Shu, A., Chen, Y., Chen, J., Yang, Y., Tang, W., & Zhang, Y. (2023). Study of convolutional neural network-based semantic segmentation methods on edge intelligence devices for field agricultural robot navigation line extraction. Computers and Electronics in Agriculture, 209, 107811. https://doi.org/10.1016/j.compag.2023.107811
Joseph, D.S., Pawar, P.M. & Pramanik, R. Intelligent plant disease diagnosis using convolutional neural network: a review. Multimed Tools Appl 82, 21415–21481 (2023). https://doi.org/10.1007/s11042-022-14004-6
