Machine Learning Based Crop Classification with Sentinel-1 Data

Abstract

Agriculture has played a crucial role in developing countries where the majority of the rural population relies on it for their livelihoods. A finer-grade crop classification has become crucial in the context of precision agriculture. In recent years, the volume of open satellite data has grown significantly, and at the same time, the cost of high-speed cloud computing has continued to decline. This can be used in combination with machine learning techniques to classify crop types in the agricultural industry. This study focuses on different machine learning algorithms for crop classification for the region of interest (ROI) in the study area of Kendarapara district, Odisha, for the year 2021 utilizing Sentinel-1 data. The study was performed using Google Earth Engine. The performance of four machine learning techniques Random Forest (RF), Classification and Regression Trees(CART), Gradient Boosting, and Support Vector Machine(SVM) algorithm, for three different crop type classifications, were evaluated. The results demonstrated that CART has the highest accuracy of 98.77%. Consequently, the method offers a straightforward yet accurate and efficient technique to classify crops.