An alert based perspective for disease detection in tomato leaves using machine learning
Loading...
Date
2023-09-01
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
G.B. Pant University of Agriculture and Technology, Pantnagar-263145
Abstract
Tomato farming plays a crucial role in ensuring food security and agricultural sustainability worldwide. However, various diseases often afflict tomato plants, posing a significant threat to crop yields and quality. Timely disease detection and intervention are essential for mitigating these threats and ensuring a bountiful harvest. In recent years, advancements in machine learning and computer vision have offered promising solutions for automating disease detection in crops. This study leverages the power of Convolutional Neural Networks (CNNs) to develop a robust and efficient system for disease detection in tomato leaves. The project employs a dataset sourced from Kaggle, comprising ten distinct classes of tomato leaf diseases, providing a comprehensive scope for disease identification. One of the critical components of this research is the use of Google Colab, a cloud-based platform that facilitates collaborative coding and GPU acceleration, making it ideal for training deep learning models. To achieve this, the study explores various training strategies, experimenting with different numbers of training epochs.
The training process involves adjusting the learning rate, a critical hyperparameter that influences the convergence and accuracy of the model. Three training scenarios are examined: 50 epochs, 100 epochs, and 200 epochs, with a fixed learning rate of 0.01. The results obtained from the experiments are highly encouraging. After 50 epochs of training, the CNN model achieves an accuracy of 83.33%. Continuing the training to 100 epochs significantly enhances the model's performance, yielding an accuracy of 94.67%, when training is extended to 200 epochs, the accuracy remains competitive at 94.50%, demonstrating that the model has reached a stable state of performance. Beyond the development of an accurate disease detection model, the study integrates Telegram, a popular messaging platform, as a means to disseminate alerts to farmers. When the CNN model identifies a diseased tomato leaf in an image, it sends an immediate alert to farmers via Telegram, providing them with essential information about the detected disease.
This research showcases the potential of CNNs and machine learning in revolutionizing agriculture by automating disease detection in tomato leaves. This project underscores the potential for technology-driven solutions to address critical challenges in agriculture, offering a pathway to greater food security and sustainability in a rapidly changing world. As agriculture continues to evolve, embracing such innovations becomes imperative for the industry's growth and resilience.
(
Description
Keywords
Citation
Academic Research