Plant Health Prediction and Monitoring Based on convolution Neural Network in North-East India

Abstract

Agriculture is the main backbone of any society. In this modern era as the population continuously increases, resources for farming are also decreasing. If the health condition of the plant’s determined at a regular interval, farmers can take action timely to improve the plant’s health condition. Plant monitoring and plant health status at regular intervals may lead to better growth of foods. But the regular physical visit to the crop field to monitor plants and plant health is a critical task for a large crop field. To overcome this situation, we need to shift from a traditional cropping system to smart agriculture. Now, these days, a smart agriculture-based approach can use the internet of things, machine learning, and deep learning to predict the health condition of plants. In this paper, the internet of things (IoT) based smart agriculture system has been presented along with machine learning, deep learning, and transfer learning to monitor and health prediction of plants. The IoT-based system has been used to monitor the plant’s surrounding parameters such as humidity, temperature, light intensity, and soil water moisture. The leaf images of plants have been used in deep learning (DL), machine learning (ML), and transfer learning (TL) to predict the health condition of plants. In this paper, the convolution neural network (CNN) based model has been proposed and it has been compared with the support vector machine (SVM), random forest (RF), VGG19, and mobilenet model. It has been concluded that the accuracy of the proposed CNN model is 81.5 %, which is the highest among SVM, RF, VGG19, and mobilenet..