AN AUTOMATAED DEEP LEARNING MODEL TO CLASSIFY DISEASES IN AREACANUT PLANT

Abstract

Detecting diseases in plant at an early stage is important for ensuring healthy crops and reducing economic losses. Traditional methods are slow and require expertise. The recent technological developments bring in a lot of computational techniques that enables the detection of diseases at an early stage and more accurate. The proposed work has been implemented using deep learning algorithms The work focuses on identifying the diseases in Arecanut leaf and analyzing the efficiency of the deep learning techniques in detecting the type of diseases. Different CNN algorithms like ReNet, MobiNet and VGG Net have been implemented and tested for thier efficiency. The appropriate model is then optimized and deployed in an Android device so as to enable the farmer to use the application efficiently. The proposed work is implemented by collecting a dataset of arecanut diseased leaf images and dividing it for training, validation, and testing. The performance of the models are compared using the parameters (trainable and non-trainable) and the utilisation of the memory during runtime. The models are evaluated based on accuracy and precision. For the given dataset, ResNet performed with 79% accuracy, MobiNet with 86% and VGG with 92% accuracy. The performance efficiency of VGGNet is better than the other two architectures and deployed in Android device to help the stakeholders.