Unmanned Aerial Vehicle for Precision Agriculture in a Modular Approach

Abstract

Agriculture plays a vital role in the economy of any country. It is an important contributor to a nation's GDP. Farmers face three main challenges to their crop produce, i.e. crop health, damages from insects, and weeds. The paper discusses an engineering solution using a UAV (unmanned aerial vehicle) in a modular approach for crop health analysis and detects the weeds and insects. A market-ready concept of low-cost UAV design with two different modules - one for crop health and another for anomaly detection presented. The proposed method facilitates the detection of weed and insects in a farm using deep learning algorithm faster convolutional neural networks (Faster R-CNN_inception_resnet) implemented using Tensorflow API. The result shows that the design of UAV was very efficient in capturing the relevant data and the model proposed for prediction came out with an improved accuracy rate of 89.5%.