With the rapid growth of live streaming, monitoring, medical imaging and video gaming services, the limited native resolution of front-end capture equipment and distortion generated during network transmission lead to widespread low-definition video issues. Traditional offline super image resolution processing cannot meet low-latency real-time service demands. Super-resolution refers to the recovery of the high-resolution (HR) image/video containing high-frequency detail information from low-resolution (LR) images/video. This Recommendation specifies the requirements for the real-time super-resolution service based on artificial intelligence (AI).