Abstract: This paper presents the results of a systematic study on infrared and visible image fusion (IVIF) technology based on deep learning algorithms. First, it comprehensively elaborates on the research status of mainstream deep learning models used for IVIF tasks; discusses the results of an in-depth analysis of the framework structures, core characteristics, and computational complexities of various methods; and reveals the advantages and disadvantages of different models for information extraction and reconstruction. Second, it systematically sorts the subjective and objective evaluation systems for IVIF and analyzes the focus dimensions of the subjective evaluation and applicable scenarios of objective evaluation metrics in detail. Third, the currently publicly available and commonly used IVIF datasets are summarized to provide a reference for algorithm verification and comparison. Finally, aiming at the shortcomings of the existing methods, future development trends for IVIF in the directions of model lightweighting, registration adaptability, and task-driven fusion are discussed.