Automated Tomato Leaf Disease Identification via Transfer Learning Approach on Convolutional Neural Network

Tomatoes are the most nutrient-dense, most cultivated, and most consumed vegetables in the world, and they are attractive due to their therapeutic properties. On the other hand, tomato crops are susceptible to numerous diseases that can lower or completely destroy tomato production; hence, the fast and precise diagnosis of tomato infections plays a vital role in crop protection. As a result, researchers have created numerous deep learning models for the automated classification of tomato leaf diseases. Transfer learning has made it possible to apply solutions from one area to another, enabling the development of stronger and more effective solutions for the diagnosis of tomato leaf diseases. This article utilized transfer learning on different forms of standard CNN architecture, like DarkNet-19, ShuffleNet, and ResNet-18, to diagnose tomato plant diseases from its leaf images. Also, this article assesses the model's performance with stochastic gradient descent (SGD), RMSProp, and adaptive moment estimation (Adam). The outcome of the experiment shows how successful the model is classifying tomato leaf diseases automatically with the transfer learning technique.