AI-Driven Agro-Advisor for Automated Tomato Disease Diagnosis and Sustainable Organic Remediation

Abstract

The agricultural industry is a major aspect of many developing countries, as substantial portions of these countries rely on agriculture as their income source. This industry faces many obstacles, including issues related to plant diseases—which create major issues for agricultural production no matter where it occurs across the globe. Therefore, to achieve sustainability in agriculture and minimize the losses caused by plant disease, it is important to develop systems that are able to identify plant diseases at an early stage, as well as to accurately predict how soon those plant diseases can impact crop yields or quality and result in loss of income to farmers. In response to this critical need, we are proposing a new enhanced AI-based agro advisory system designed for automated automated identification of tomato plant diseases leveraging advanced deep learning techniques.The agro advisory system leverages the MobileNetV2 convolutional neural network architecture to create a lightweight/efficient model to classify images of healthy versus diseased tomato leaves. The model is trained using a publicly available dataset containing thousands (over 16,000) unique images of healthy versus diseased tomato leaves, providing a rich set of training data to enable the model to learn how to identify unique distinguishing characteristics between healthy and diseased tomato leaves. Farmers and other users can access the Tomato Disease Prediction (TDP) agro advisory service interface by using a Flask web-based server and uploading images of their tomato leaves; receiving real time prediction reguarding the presence of disease on their tomato plants and which specific type of tomato disease has been detected. The interactive nature of this system provides an easy-to-use, accessible solution for farmers and others with little-to-no technical expertise through an intuitive interface; thus making it a great addition to helping farmers produce safe and bountiful foods to feed the world's hungry. This work contains an important area of development with the addition of scalable database solutions, MySQL and MongoDB, which can be used to manage user data, historical predictions and disease-specific data effectively; thus enhancing both the scalability of the system and its ability to perform. Compared with traditional methods, the proposed system provides a more accurate prediction of disease occurrence, a quicker response time and improved data management. Overall, this AI-based platform presents a useful, economical, dependable means for agricultural production by limiting the reliance of the agricultural community on human efforts to diagnose a plant's disease and thus helping with proper decision making about crop production and ultimately increasing agricultural productivity.