Rice Leaf Disease Identification Using Deep Learning

Abstract

We have developed an online app to identify rice leaf diseases automatically through deep learning models. The gravest challenges to crop production and food safety are posed by rice leaf diseases like Tungro, Blast, Bacterial Blight, and Brown Spot. Several models were used in this study such as MobileNetV2 individually, ResNet50V2 separately, and DenseNet201 on its own as well as a combination model which is made up of ResNet50V2 and DenseNet201 known as the hybrid model. Following the validation process the hybrid model achieved an accuracy rate of up to 99.83% with 99.77% during training while MobileNetV2 got 98.80% in the validation stage after being trained at 98.30%. ResNet50V2 reached 98.30% on validation after hitting 99.00% during training while DenseNet201 had 98.00% after 96.30% for training. The user-oriented design lets people upload images and choose a classification model. This gives quick and correct results for real-time use. The approach makes it easier to identify diseases and helps agricultural professionals and farmers access them easily so that they can be managed promptly and effectively. To solve this problem, we have brought these models onto a platform where they are easy to reach. This is expected to reduce the effects of leaf diseases on rice significantly hence promoting sustainable farming as well as food security.