1Department of Computer Applications, Kasegaon Education Society’s, Rajarambapu Institute of Technology, affiliated to Shivaji University, Sakharale-415 414, Maharashtra, India
2KL Business School, Koneru Lakshmaiah Education Foundation, Guntur-521 180, Andhra Pradesh, India
3Department of Language, Culture and Society, SRM Institute of Science and Technology, Delhi NCR Campus, Modinagar, Ghaziabad-201 204, Uttar Pradesh, India
4Department of Computer Science and Business Systems, Dayananda Sagar College of Engineering, Bangalore-560 111, Karnataka, India
*Corresponding Author: Namrata A. Jadhav, Department of Computer Applications, Kasegaon Education Society’s, Rajarambapu Institute of Technology, affiliated to Shivaji University, Sakharale415 414, Maharashtra, India. Email: namratakoravi09@gmail.com
The timely and precise detection of foliage diseases is essential for the efficient management of bean crops. Conventional techniques for detecting leaf diseases need professionals to physically examine the leaves, which is laborious and subject to human mistakes. Due to advanced machine learning (ML) computations, convolutional neural networks (CNNs) are capable of examining leaf diseases. MobileNetV3 is particularly effective for edge device applications because of its lightweight and efficient design.
In this study, the effectiveness of MobileNetV3 in distinguishing three disease categories: Angular leaf spot, bacterial blight and healthy, from images of bean leaves is evaluated. The dataset utilized in this study is sourced from Kaggle and comprises a total of 1295 images categorized into three classes. Each image was resized to 224×224 pixels and normalized to a range of [0, 1] to prepare it for input into the MobileNetV3 model. Data augmentation techniques, including rotation, flipping and zooming, were applied to enhance the model’s generalization capability. The MobileNetV3 architecture was implemented with a batch size of 32 and trained for 25 epochs using the Adam optimizer with a learning rate of 1e-4. Performance was evaluated based on accuracy, precision, recall and F1-score, using a split dataset consisting of training (80%), validation (10%) and test (10%) subsets.
The MobileNetV3 model demonstrated a high level of performance in classifying bean leaf diseases. The model achieved an overall accuracy of 92.19% with a macro average F1-score of 92.27%. These results indicate that MobileNetV3 is effective for automated classification of bean leaf diseases, with high precision and recall, particularly for angular leaf spot.
Agricultural technology, Convolutional neural networks, Data augmentation, Image classification, MobileNetV3
