Mobile sensing for agriculture activities detection

Abstract

The agriculture activities have a major role in determining the quality and quantity of the agriculture produce. In this paper, we propose a novel mobile sensing based framework which uses machine learning algorithms for the detection of agriculture activities. To collect the sensors data and ground truth an android based mobile application has also been developed and has been provided to the farmers. We investigate the performance of Naive Bayes, Linear Discriminant Analysis (LDA) and k-Nearest Neighbor (k-NN) classifiers to detect the activities like Harvesting, Bed Making, Stand-still and Walking. We also use the same classifiers to detect the placement of the mobile phone on the body which will hence provide a degree of freedom to the farmers in placing the mobile phone as per their convenience.