Publication

Abstract : The use of remote sensing in smart farming is gaining popularity around the world. Due to the rising global demand for r food grains, assessing yield before actual production is critical in developing policies and making decisions in the agricultural production system. In this project, the aim is to develop a deep learning model which will predict an individual farm’s crop yield using remotely sensed satellite images. The study focuses on crop yield estimation of wheat, corn and soybean in the chosen study area in KBS-LIER, Michigan, United States. Four vegetative spectral indices, including NDVI and SAVI, etc. were calculated from the seven spectral bands collected over 11 years. Model-I, trained on spectral bands data, achieves an accuracy of 94.3%, outperforming Model-II, which utilizes vegetative indices data and achieves and accuracy of 88.8%]. ©2024 IEEE.