Land use land cover classification of remote sensing images based on the deep learning approaches: a statistical analysis and review

Over the last few years, deep learning (DL) techniques have gained popularity and have become the new standard for data processing in remote sensing analysis. Deep learning architectures have drawn significant attention due to their improved performance in a variety of segmentation, classification, and other machine vision applications. In remote sensing, land use and land cover (LULC) are critical components of a wide variety of environmental applications. Changes in land use on a spatial and temporal scale occur due to accuracy, the capacity to develop, flexibility, uncertainty, structure, and the capability to integrate available models. Therefore, LULC modeling’s high performance demands the employment of a wide variety of model types in remote sensing, which include dynamic, statistical, and DL models. In this study, we first analysed several key findings and research gaps in traditional technology while discussing various software applications used for LULC analysis. Second, the fundamental DL and ML concepts applicable to LULC are introduced with their merits and demerits. We employ a comprehensive review of distinct DL architectures and a custom framework to handle the challenging task of detecting changes in LULC. Subsequently, a detailed statistical analysis is conducted on the”Scopus database” to ascertain current trends in LULC utilising DL methods. This overview encompasses practically all applications and technologies in the field of LULC, from preprocessing to mapping. Finally, we conclude with a proposal for researchers to perform future potential using state-of-the-art methodologies.