Real-time AI super resolution method and system for game interested target and computer readable storage medium

Abstract

The invention provides a real-time AI super resolution method and system for a game interested target and a computer readable storage medium. Firstly, the overall resolution of a game picture is reduced, and the image rendering degree of an intelligent terminal is reduced, so that the power consumption is reduced to a certain extent; the game picture with low resolution is identified, a plurality of interested targets are obtained, computing power of different sizes is distributed to the interested targets with different priorities to achieve different definitions, so that the interested targets with higher priorities are higher in definition, and the interested targets with lower priorities can be lower in definition; and the AI computing power of a plurality of interested targets with lower priorities is reduced, the computing speed is integrally improved, the performance is further improved, and the temperature power consumption of the intelligent terminal is reduced.

Description

technical field [0001] The present invention relates to the technical field of game screen display, in particular to a real-time AI super-resolution method, system and computer-readable storage medium for objects of interest in games. Background technique [0002] With the continuous improvement of AI computing power of smart terminals, it is expected to move the work of GPU to AI. However, on smart terminals, AI super-resolution is often used for video streaming, and the requirements for delay are low, and the video has a clear current frame and future frame, while the game is rendered in real time, and no future frame is used, only the past frame and the current frame At the same time, due to the need to calculate the precise motion vector of each pixel, the HDR color, the precise depth of the pixel, etc., its calculation amount is relatively large, which increases the power consumption and the delay of the picture output. Contents of the invention [0003] In order to ...

AI Technical Summary

Problems solved by technology

However, on smart terminals, AI super-resolution is often used for video streaming, and the requirements for delay are low, and the video has a clear current frame and future frame, while the game is rendered in real time, and no future frame is used, only the past frame and the current frame At the same time, due to the need to calculate the precise motion vector (motion vector) of each pixel, the HDR color, the precise depth of the pixel, etc., its calculation amount is relatively large, which increases the power consumption and the delay of the picture output

Images