Abstract: Underwater images are important carrier and representation of underwater information, which is crucial for ocean resources’ exploration, development, and utilization. However, due to the limitations of objective imaging environment and imaging equipment, the quality of underwater image is always unsatisfactory, such as low contrast, blurred details, and color deviations, seriously restricting the development of related research areas. Therefore, how to enhance and restore degraded underwater images through the designed post-processing algorithms has attracted more and more attention. In recent years, with the rapid development of deep learning strategy, deep learning-based underwater image enhancement and restoration technology has also made great progresses. In order to comprehensively summarize the existing methods and keep up with the latest research progress, this paper introduces the methods and models of deep learning driven underwater image enhancement and restoration technology, summarizes the existing underwater image datasets in detail, analyzes the key issues of the existing deep learning-based methods, and prospects for the future development directions.