Soil Nutrients Analysis Using CNN Over Linear Regression

Abstract

Traditional inspection and laboratory testing methods of soil nutrient analysis are often expensive and excessively rely on human labour. This study focuses on analyzing soil nutrients through structured datasets with two models, Convolutional Neural Networks (CNN) and Linear Regression, without involving image processing. In our proposed system, soil nutrient datasets, which contain data like pH, moisture, and nutrient levels, are ingested, Preprocessed, feature extracted, and the two models are trained. Each model is evaluated and compared in terms of their predictive accuracy, and in this study, it is shown that CNN outperforms in capturing intricate patterns in the data, while linear regression is appropriate for less complicated situations. With this method, nutrients in the soil can be predicted in real time, facilitating intelligent decisions for precision agriculture. Further improvement of the model concerning accuracy, scalability, and efficiency for general agriculture purposes remains a work in progress