A method and system for generating high frame rate video based on deep learning

Abstract

The invention discloses a method for generating high frame rate video based on deep learning, comprising: using one or more original high frame rate video clips to generate a training sample set; using multiple video frame subsets in the training sample set to train dual Passage convolutional neural network model, to obtain optimized dual-channel convolutional neural network, the dual-channel convolutional neural network model is a convolutional neural network fused by two convolution channels; using the optimized dual-channel The convolutional neural network generates an interpolation frame of any two adjacent video frames in the low frame rate video, so as to generate a video with a higher frame rate than the low frame rate video. The whole process of the method of the present invention is end-to-end, without subsequent processing of video frames, the effect of video frame rate conversion is good, the fluency of the synthesized video is high, and the problem of jitter, video scene switching and other problems existing in the video shooting process is relatively good. Good robustness.

Description

technical field [0001] The invention belongs to the technical field of computer vision, and more specifically, relates to a method and system for generating high frame rate video based on deep learning. Background technique [0002] With the development of technology, it is more and more convenient for people to obtain videos. However, due to hardware reasons, most videos are collected by non-professional equipment, and the frame rate is generally only 24fps-30fps. High frame rate video has extremely high fluency, which can bring people a better visual experience. If people directly upload high frame rate videos to the Internet, due to the increase in traffic consumption, people's costs will also increase. If the video with a low frame rate is directly uploaded, due to the reasons of the network line, there will inevitably be a problem of frame loss during the video transmission process. The larger the video, the more likely this phenomenon will occur, so that the remote vi...

AI Technical Summary

Problems solved by technology

[0003] Aiming at the above defects or improvement needs of the prior art, the present invention provides a method for generating high frame rate video based on deep learning, the purpose of which is to convert low frame rate video into high frame rate video, thereby solving the problem caused by low frame rate Frame rate video frame loss during network transmission, resulting in a decrease in video quality and a technical problem that affects people's experience

Images