Review of Remote Sensing Image Classification: Technology Evolution, Method Innovation and Future Challenges

Remote sensing image classification is the core technology of environmental monitoring, urban planning, and resource management. Its goal is to accurately identify the ground cover through intelligent algorithms. In recent years, breakthroughs in deep learning technologies (such as CNN and Transformer) have significantly improved classification accuracy and efficiency, but problems such as generalization in complex scenes, high annotation costs, and insufficient real-time performance remain to be addressed. In this paper, the theoretical basis and technical evolution of remote sensing image classification are systematically reviewed. By combining with the latest research progress from 2024 to 2025, the research directions of multimodal fusion, lightweight networks, small sample learning etc. are analyzed, and the future development trends of self-supervised learning, edge computing and so on are discussed. By integrating classical methods and intelligent technologies, this paper aims to provide a comprehensive technical reference and development directions for researchers in the field of remote sensing classification.