沈世银，中国科学院上海天文台研究员，主要从事星系结构和演化研究，任载人空间站巡天望远镜（CSST）科学数据系统副主任设计师；两次获得上海市自然科学二等奖，入选上海市优秀学术带头人，2021-2024爱思唯尔中国高引学者。In recent years, with the development of AI, the application of computer vision technology (CV) in astronomical images has begun to accelerate. With the advent of the next generation of astronomical image surveys (e.g. CSST), the application and demand for AI technic may give rise to the "AlphaFold moment" in astronomy. In this talk, I will first introduce the differences between astronomical images and traditional CV images: low signal-to-noise ratio, errors, and masks. It is these features that induces the huge demand and challenges for the AI applications on astronomical images. We started from the requirements of astronomical study and leveraged the advantages of AI technic in computational efficiency and image feature extraction to conduct a series of applications from galaxy image data (A) to requirements (B), demonstrating the huge potential of AI technic in astronomy. Finally, I will further look ahead to the needs of AI technic in astronomical image studies.