Satellite Images Analysis and Classification using Deep Learning-based Vision Transformer Model

Analysis and classification of satellite images from diverse sources, including remote sensing and satellite devices, have been explored to understand the dynamics of land use. However, due to their high complexity and multi-resolution, multi-spectra, and multi -scale nature, traditional machine learning classifiers have limitations in their analysis. In this research, an advanced machine learning technique, a deep learning-based vision transformer model, which leverages the benefits of selfattention mechanisms to overcome the challenges of analyzing complex features in satellite images, is proposed for efficient classification. Experimental evaluation on the publicly available EuroSAT satellite imagery dataset demonstrates promising results, achieving an accuracy of 98%.