Leaf Disease Detection by Using Convolutional Pretrained Model

Abstract

Although agriculture plays an important role in developing countries such as India, food security remains a major concern. Due to a shortage of storage space, transportation, and plant diseases, the majority of crops are squandered. In India, infections cause more than 15% of crops to be wasted, making it one of the most pressing issues to be addressed. There is a need for an autonomous system that can detect these illnesses and assist farmers in taking the necessary procedures to avoid crop loss. Farmers have used the traditional approach of recognizing plant illnesses with their naked eyes. However, not all farmers can recognize these diseases in the same way. With the advancement of Artificial Intelligence, there is a need to apply computer vision capabilities to the agricultural area. Deep Learning's comprehensive libraries, as well as the user and developer-friendly environment in which to work, all combine to make Deep Learning the best way to get started with this topic. Taking leaves from diseased crops and identifying them according to the disease pattern is part of the process. Images of diseased leaves are processed using pixel-based procedures to improve the informational content of the images. The next step is feature extraction, image segmentation, and finally, classification of crop diseases based on patterns recovered from diseased leaves. Convolutional Neural Networks (CNNs) are used to classify diseases. Some of the deep learning pre-trained models have got more accuracy here. The comparison of two pre-trained models was shown.