Enhanced Agricultural Monitoring Through Hyperspectral Imaging and Advanced Machine Learning Techniques
Nellore Kapileswar, Judy Simon, Kota Sirisha, Bezawada Raja Pujitha, Lekkala Charan Sai Kumar, Chappagadda Harish
Abstract
Agriculture has many challenges due to climate change, limited arable land and population growth. Hyperspectral image (HIS) provides an efficient solution by monitoring the crop health and identifying the disease utilizing the spectral information. In this study, a machine learning framework for hyperspectral image analysis for enhancing the agricultural monitoring. The framework incorporates various pre-processing steps, including spectral data calibration, normalization, and dimensionality reduction, to prepare the hyperspectral data for analysis. Feature extraction techniques, such as spectral signature analysis is employed to identify key indicators of crop health. Also, the 3D-CNN and Recurrent Neural network (RNN) are implemented to obtain the spatial and spectral patterns. Experimental analysis is conducted using 5-fold cross validation with relevant performance metrics such as precision, accuracy, F1-score and recall. This study benefits farmers with correct and precise advancements in crop health. It also enhances the yield and eliminating the losses.