An Efficient Regression Method To Predict Soil pH Using RGB Values

Abstract

The fertility of a soil is governed by potential of Hydrogen (pH) value of the soil. This research paper presents a novel approach for predicting the pH value of a soil by using RGB (Red, Green, Blue) values of an image. The study uti- lizes machine learning techniques to develop a model that can accurately pre- dict the soil pH based on the colour information captured in an image of the soil. The model was trained with a dataset containing RGB and correspond- ing pH value as the attributes and tested using a variety of images. Results show that the proposed model is able to predict soil pH with minimal error, demonstrating the potential for using image analysis as a practical and effi- cient method for soil pH determination in agriculture and soil science. With the available dataset, various regression approaches have been implemented to predict the soil pH value, and eventually the experimental results shows that the polynomial regression is the most effective method as the data is not linear for analysing this dataset.