EAgri: Smart Agriculture Monitoring Scheme using Machine Learning Strategies

Abstract

The logic of Machine Learning and its predictive strategies are applied to many different applications to attain good benefits over now-a-days. This paper associates the machine learning concept to improve the production on agricultural field as well as the novel adaptive technologies are associated into this learning concept to make a proper agricultural monitoring system in fine manner. This paper is intended to design a new Agricultural Monitoring robot called eAgriBot, in which it integrates the logic of Machine Learning and produce an intelligent predictions to prevent the crops from affections including weather conditions, rainfall and soil water level. In parallel, the eAgriBot contains a high resolution digital camera to capture the pictures of the crops and maintains that into the server unit in proper manner. In literature, there are many approaches designed to provide an automated watering system, systematic pesticide spraying and so on. But all are dependent on the human operations, in which the automatic watering system requires the manual trigger from either SMS or other internet associated operations; similarly the systematic pesticide mechanism requires the same kind of trigger to perform the action. These cases are critical in terms of monitoring the agricultural field from remote environment. The concept of Internet of Things (IoT) is associated over this approach to push and update the agricultural data collected by the eAgriBot to the Cloud Server. This entire process is controlled and manipulated by the novel machine learning strategy called Smart Learning Assisted Data Manipulation (SLADM), in which it is derived from the traditional Random Forest Classification logic with specific parameter modification called dynamic threshold fixation. In general the Random Forest logic uses the constant threshold for data processing, but in this approach dynamic principles are applied to improve the prediction accuracy. With the help of this system plants leaf disease can easily be monitored with the help of digital camera associated with the eAgriBot. It captures the crop field images and pass it to the server end for processing, in which the server end accumulates that and process that by using proposed machine learning called Smart Learning Assisted Data Manipulation. The proposed SLADM scheme is designed robustly for analyzing the both the image and data content, so that the logic identifies the severity in leaves and the data as well. In case of any severity level mismatching found by the algorithm, it immediately notifies that to the respective farmer to take an appropriate action to save the plants from the disease spread. The data available into the cloud server can easily monitored by the farmers from anywhere in the globe as well as this approach of SLADM provides an accurate agricultural field predictions, so that the farmers can easily monitor the field without any complexities and attain good production level easily