Plant Disease Detection Using Transfer Learning with DL Model

Abstract

Crop disease is a huge danger to food security, but due to a lack of infrastructure in many regions of the world, timely identification is challenging. The most popular method for identifying those infections is to see the plant with the naked eye; however, this method does not provide an accurate solution and takes a long time. As a result, applying deep learning techniques such as transfer learning to improve plant disease detection could be beneficial. ResNet50 with transfer learning is used to identify plant diseases in this study. ResNet50's performance is compared to VGG-16 and Inception V3, which were created and trained from the ground up. ResNet50 had the maximum performance of 96.23%, and its performance was 95.06% after using the tenfold cross-validation procedure to validate the outcome. The model can identify 38 different types of plant diseases.