Comparative study using different convolutional neural network models to predict leaf diseases in plants

Abstract

The four common types of leaf diseases are Rust, Scab, Multiple diseases, Healthy. Effects of certain bacteria, micro-organisms and fungi affect the growth and development of leaves which can be stopped by early detection and accurate identification of leaf diseases and can also insure less spreading of infection and a healthy development of leaf takes place. This research paper use image pre-processing and can generate high recognition rates for leaf diseases. A dataset of 3642 images is taken and trained by different models like VGG16, ResNet50, InceptionV3, InceptionResNetV2 with the help of deep learning algorithm like convolutional neural networks and transfer learning approach for real time detection of leaf diseases. By training the leaves based on the proposed models we will be able to know the diseases present in the leaves. The purpose of this research paper is based on the comparison of accuracy given by different models when they are trained.