Unveiling Soil Fertility Patterns via Image Analysis and Machine Learning for Accurate Crop Recommendations

Abstract

Accurate fertilizer use is essential for precision agriculture. Crop recommendation systems should consider real-time data on soil fertility, crop type, soil nutrients, and environment to ensure sustainability and profitability. Due to its intricacy, real-time fertility of soil mapping is expensive, time-consuming, and costly. Using the Image's real-time soil fertility mapping data, we offer a crop selection approach based on artificial intelligence. It is suggested that an image architecture aid soil fertilizer mapping. Real agricultural fields have been employed to assess the precision of IMAGE-based fertility mapping employing the provided method. When evaluating IMAGE base fertility mapping compared to the process of soil chemical analysis, we look at its ability to accurately track nitrogen (N), phosphorus (P), potassium (K), and other environmental variables including humidity, precipitation, temperatures, and pH. Based on the soil fertilizer information, the following machine learning techniques are used to suggest crops: Logistic Regression (LR), Support Vector Machine (SVM), K-Nearest Neighbour (KNN), Random Forest (RF), Decision Tree (DT), and XGBoost (XGB). All machine learning algorithm performance is well good but the GNB and RF is performed most accurate performance as compared to other machine learning modules.