Comparative Analysis of Machine Learning and Deep Learning Models for LULC Classification using Remote Sensing Data

Objectives: The primary objective of this study is to evaluate and compare the performance of machine learning and deep learning models for Land Use and Land Cover (LULC) classification using remote sensing data. Specifically, it assesses Support Vector Machine (SVM), XGBoost, an ensemble model (SVM + XGBoost), and a Deep Neural Network (DNN) on pre-processed hyperspectral datasets (Pavia University, Indian Pines) and raw satellite imagery from the Twin Cities of Odisha, India. Method: The study follows a systematic workflow, including data acquisition, preprocessing using Principal Component Analysis (PCA), and classification using machine learning and deep learning models. The models are evaluated based on Overall Accuracy (OA), Kappa Index, Precision, Recall, and F1-score. Statistical significance is tested using the Nemenyi post hoc test to determine the superiority of the proposed ensemble model over individual classifiers. Findings: The ensemble model (SVM + XGBoost) achieved the highest classification accuracy across all datasets: 95.4% for Pavia University, 88.7% for Indian Pines, and 93.6% for the Twin Cities of Odisha. The statistical analysis confirms that the ensemble m