Decision tree based mobile crowdsourcing for agriculture advisory system

Abstract

In agriculture, the major cause of loss of yield and quality of harvest is due to the outbreak of pest and diseases. It is difficult for fewer experts to visit a large number of farms to analyze the pest and diseases and to provide alerts to other farmers in the same region. Therefore, remote classification of pest and diseases is essential to advise on appropriate corrective actions to the farmers. In this paper, we propose a mobile phone based crowdsourcing approach to detect the emergence of plant diseases in a region by aggregating information from multiple farmers. The proposed crowdsourcing approach is based on the binary decision tree that helps to minimize the number of questions to be asked to the farmers and results in an accurate and reliable decisions on plant diseases. The results are presented by using the soybean disease large dataset from the UCI machine learning repository for the classification of 15 prevalent diseases in the soybean crop.