Comparative analysis of tiny machine learning models for maize crop disease identification
Keywords:
CNN, deep neural network, disease identification, EfficientNet, maize crop, MobileNet, SqueezeNet, ShuffleNet, Tiny Machine LearningAbstract
This paper presents a comprehensive analysis of tiny machine-learning models for identifying maize crop diseases. The study evaluates custom deep neural network (DNN) models alongside popular deep learning architectures in terms of their effectiveness and efficiency in disease identification. The traditional crop disease detection methods such as manual checking of crop leaves for defects have become obsolete hence the need for modern and machine learning methods that are accurate and precise. This paper uses a dataset with maize crops as an example. This paper also compares four types of deep learning architectures used with custom deep learning architectures namely MobileNet, EfficientNet, ShuffleNet, and SqueezeNet. MobileNet is a lightweight CNN for mobile devices, it uses depth-wise separable convolutions for efficiency. EfficientNet is a scalable CNN architecture balancing depth, width, and resolution for optimal performance across resource-constrained devices. ShuffleNet is an optimized CNN with channel shuffle operations for cross-group information flow, ideal for resource-constrained environments. SqueezeNet is a compact CNN-based deep learning architecture utilizing fire modules to minimize parameters while preserving accuracy, it is suitable for IoT and embedded systems. The comparison of these architectures was done using the accuracy, loss, train time, and validations as the model selection criteria for comparison. The results show that MobileNet and SqueezeNet outperform both the Custom Model and ShuffleNet in terms of both test loss and test accuracy. EfficientNet, however, shows significantly poorer performance compared to the other models, particularly in terms of test accuracy. The Custom Model achieved a test loss of 0.15 with a test accuracy of .96.38%, while MobileNet attained a lower test loss of 0.098 and a higher test accuracy of 97.32%. In contrast, EfficientNet exhibited the highest test loss of 2.79 with the lowest test accuracy of 28.43%. ShuffleNet showed a test loss of 0.12 and a test accuracy of 96.18%, whereas SqueezeNet achieved a slightly lower test loss of 0.09 but a similar test accuracy of 96.84. Contribute to advancing the development of precision agriculture technologies tailored to address crop health challenges.
References
Aggarwal, S., Sahoo, A.K., Bansal, C. & Sarangi, P.K. (2023) Image classification using deep learning: A comparative study of VGG-16, InceptionV3, and EfficientNet B7 models. 1728–1732. https://doi.org/10.1109/icacite57410.2023.10183255.
Arun, Y. & Viknesh, G.S. (2022) Leaf classification for plant recognition using EfficientNet architecture. Proceedings of IEEE 2022 4th International conference on advances in electronics, computers and communications, ICAECC 2022. https://doi.org/10.1109/ICAECC54045.2022.9716637.
Chen, C., Zhu, T., Li, S. & Liu, B. (2021) Apple leaf disease recognition method base on improved ShuffleNet V2. Proceedings - 2021 3rd International conference on advances in computer technology, information science and communication, CTISC 2021, 276–282. https://doi.org/10.1109/CTISC52352.2021.00057.
Chen, J. & Ran, X. (2019) Deep learning with edge computing: A review. In proceedings of the IEEE (Vol. 107(8), (1655–1674). Institute of electrical and electronics engineers Inc. https://doi.org/10.1109/JPROC.2019.2921977.
Chen, J. & Zhang, D., Suzauddola, M., Nanehkaran, Y.A. & Sun, Y. (2021) Identification of plant disease images via a squeeze-and-excitation MobileNet model and twice transfer learning. IET Image Processing, 15(5), 1115–1127. https://doi.org/10.1049/ipr2.12090.
Da Silva, J., Flores, T., Júnior, S. & Silva, I. (2023) TinyML-Based pothole detection: A comparative analysis of YOLO and FOMO model performance. 2023 IEEE Latin American conference on computational intelligence (LA-CCI), 1–6. https://doi.org/10.1109/LA-CCI58595.2023.10409357.
Dey, D. (2020) Jadavpur University. Electrical engineering department, institute of electrical and electronics engineers. Kolkata section. & IEEE signal processing society. Kolkata chapter. (n.d.). Proceedings of 2020 IEEE applied signal processing conference (ASPCON): Oct 7-9, 2020.
Hassan, S.M., Maji, A.K., Jasiński, M., Leonowicz, Z. & Jasińska, E. (2021) Identification of plant-leaf diseases using cnn and transfer-learning approach. Electronics (Switzerland), 10(12). https://doi.org/10.3390/electronics10121388.
Hidayatuloh, A., Nursalman, M. & Nugraha, E. (n.d.) Identification of tomato plant diseases by leaf image using Squeezenet Model.
Howard, A.G., Zhu, M., Chen, B., Kalenichenko, D., Wang, W., Weyand, T., Andreetto, M. & Adam, H. (2017) MobileNets: Efficient Convolutional Neural Networks for mobile vision applications. http://arxiv.org/abs/1704.04861.
Iandola, F.N., Han, S., Moskewicz, M.W., Ashraf, K., Dally, W.J. & Keutzer, K. (2016) SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model size. http://arxiv.org/abs/1602.07360.
Kaur, G., Sharma, N. & Gupta, R. (2023) Wheat leaf disease classification using EfficientNet B3 Pre-Trained architecture. 2023 IEEE International Conference on Research Methodologies in knowledge management, Artificial Intelligence and Telecommunication Engineering, RMKMATE 2023. https://doi.org/10.1109/RMKMATE59243.2023.10369587.
Liu, Y., Zhao, Z., Zhu, J., Shen, Z. & Sun, L. (2021) A classification algorithm of grain crop image based on improved SqueezeNet model. 2021 IEEE 3rd International Conference on Frontiers Technology of Information and Computer, ICFTIC 2021, 246–252. https://doi.org/10.1109/ICFTIC54370.2021.9647085.
Nourish, A., Batra, S., Sharoon, K., Sharma, R. & Sharma, M. (2023) A study of deep learning based techniques for the detection of maize leaf disease: A short review. Proceedings of the 7th International Conference on Intelligent Computing and Control Systems, ICICCS 2023, 134–141. https://doi.org/10.1109/ICICCS56967.2023.10142559.
Panigrahi, K.P., Das, H., Sahoo, A.K. & Moharana, S.C. (2020) Maize leaf disease detection and classification using machine learning algorithms. Advances in Intelligent Systems and Computing, 1119, 659–669. https://doi.org/10.1007/978-981-15-2414-1_66.
Rahul Kumar, V.H., Shrishti, V.H. & Sridhar, P.A. (2022) Corn plant disease classification using a combination of machine learning and deep learning. 2022 International Conference on Futuristic Technologies, INCOFT 2022. https://doi.org/10.1109/INCOFT55651.2022.10094326.
Safie, S.I., Kamal, N.S.A., Yusof, E.M.M., Tohid, M.Z.W.M. & Jaafar, N.H. (2023) Comparison of SqueezeNet and DarkNet-53 based YOLO-V3 performance for beehive intelligent monitoring system. 13th IEEE Symposium on Computer Applications and Industrial Electronics, ISCAIE 2023, 62–65. https://doi.org/10.1109/ISCAIE57739.2023.10165285.
Savary, S., Ficke, A., Aubertot, J. N. & Hollier, C. (2012) Crop losses due to diseases and their implications for global food production losses and food security. Food Security (Vol. 44, (519–537). https://doi.org/10.1007/s12571-012-0200-5.
Tan, M. & Le, Q.V. (2019) EfficientNet: Rethinking model scaling for convolutional neural networks. http://arxiv.org/abs/1905.11946.
Wijaya, E.S., Mizwar, A., Islami, A.M., Sari, Y., Maulidiya, E. & Gani, I.M.A. (2022) Garbage classification using CNN architecture ShuffleNet v2. 2022 7th International Conference on Informatics and Computing, ICIC 2022. https://doi.org/10.1109/ICIC56845.2022.10006944
Yi, M., Zhao, C., Liao, F. & Yao, W. (2022) Classification of blueberry varieties based on improved EfficientNet. 2022 4th International Academic Exchange Conference on Science and Technology Innovation, IAECST 2022, 411–415. https://doi.org/10.1109/IAECST57965.2022.10062152.
Yumang, A.N., Baguisi, J.M., Buenaventura, B.R.S. & Paglinawan, C.C. (2023) Detection of black sigatoka disease on banana leaves using ShuffleNet V2 CNN architecture in comparison to SVM and KNN Techniques. 2023 15th International Conference on Computer and Automation Engineering, ICCAE 2023, 281–286. https://doi.org/10.1109/ICCAE56788.2023.10111367.
Zhang, X., Zhou, X., Lin, M. & Sun, J. (2017) ShuffleNet: An extremely efficient convolutional neural network for mobile devices. http://arxiv.org/abs/1707.01083.
