Fertilizer Spraying using Active Learning in Precision Agriculture to Improve the Crop Yield

The pests and crop detection play a major role in increasing the yield of crops, where pesticides and fertilizers help in prevention and control. The precision agriculture involving variable spraying technology is developing in recent past. However, most techniques are still in infancy level. Hence, it is essential in building a faster and efficient system in improving the spraying and plant protection mechanism. In this paper, a fertilizer spraying techniques involve with map matching algorithm using GPS is developed. In the study, a technique for map matching is proposed. (MMA) as an interactive system that enables reduced annotation cost and improved accuracy via active learning model. The models map the location points as a progression in the Hidden Markov Model (HMM) and generate automatically the results at initial iteration. The error prone points of spraying the pesticides are collected via trajectory and annotator is validated to reduce the error in finding the location points. The farmer can perform these operations either on online or in manual mode. The online model needs to set with the location details of the land to be sprayed and based on the GPS fitted in the vehicle, the fertilizer or pesticide are sprayed using the algorithm. Also, the decision of fertilizer or pesticide is selected employing a method known as map matching, which is based on the needs of the land.

I. Introduction

A major element influencing agricultural output and quality is pests and diseases, with chemical pesticides serving as the primary tool for prevention and management. In China, pesticides are currently applied using a consistent spraying technique. As a result, insecticides and herbicides are still rarely used. Statistically speaking, just 36.6% of pesticides were used in China as of 2017, which is less than half of what was used in industrialized countries.