Sensor based approach for early disease detection in plants

Abstract

Among various issues in agriculture, one is depreciation in yield of crops. India loses 15-25 % of potential crop output due to pests, weeds and diseases.This project focuses on monitoring the health of crops.It is based on the idea of giving a sensor based solution to predict the disease earlier which helps farmers to make control strategies and timely management measures to ensure stable yield of crops. Detection of the Citrus Scab, Powdery Mildew,Anthracnose and Gummosis in plants is done by prediction of parameters which determine the probability of presence of diseases. Compared different regression methods and neural network to predict Ambient temperature, Ambient humidity and Soil moisture of the soil monitoring system deployed at 3 different fields. Two different approaches are carried out for this and one is API to node data point values prediction which can predict 15-30 days earlier and other is node to node data point values prediction which can predict 4 to 8 hours earlier. Validation of these results is done by the farmers at the respective fields fields.