Towards an Effective Tomato Leaf Disease Classification Using Modified Transfer Learning Algorithm Based on resnet 50

Abstract

To increase agricultural productivity, plant diseases must be detected early and accurately. The deep learning approach based on artificial intelligence is critical in detecting illnesses utilizing a large volume of plant leaf photos. However, utilizing deep learning algorithms to identify illness with little datasets is a difficult challenge. One of the most prominent deep learning algorithms for reliably detecting plant disease with minimum plant picture data is transfer learning. This study suggests a transfer learning-based strategy for identifying tomato leaf disease. The model detects illness by combining real-time and archived photos of tomato plants. Adam, SGD, and RMSprop optimizers are also used to assess the performance of the suggested model. The experimental results show that the suggested model, which employs a transfer learning technique, is successful in classifying tomato leaf diseases automatically. When compared to SGD and RMSprop optimizers, the Adam optimizer delivers higher accuracy.