Conference Video Splitting Method and System

Abstract

The invention discloses a method and system for dismantling conference videos, wherein the method includes the following steps: acquiring a video to be processed, extracting voice text data and face data in the video to be processed, and dividing the face data according to time Mapped into the voice text data, generate voice sentence data, the voice sentence data includes time data and face identification, wherein time data includes start time stamp and / or end time stamp; Whether the identity is the speaker of the conference, obtain the judgment result; generate split point data according to the judgment result and the time data, split the video to be processed based on the split point data, and generate split fragments. The invention can automatically dismantle conference reports, save labor costs, and has high dismantling efficiency.

Description

technical field [0001] The invention relates to the field of video processing, in particular to a conference video stripping method and system. Background technique [0002] With the development of the Internet, people can watch conference videos in addition to going to the conference site; however, most conferences last for a long time, and viewers are usually only interested in some of the conference clips in the entire conference. In view of this, the industry Usually, the conference video is divided into strips so that users can quickly find the video clips they are interested in. [0003] Nowadays, there are many ways to disassemble news, such as transition recognition, voice fingerprint and face recognition; [0004] For the transition recognition technology, since the conference usually does not need to transition, it is impossible to split the conference video; [0005] For the voice fingerprint technology, because the current voice fingerprint technology is not ma...

AI Technical Summary

Problems solved by technology

[0004] For the transition recognition technology, since the conference usually does not need to transition, it is impossible to split the conference video;

[0005] For the voice fingerprint technology, because the current voice fingerprint technology is not mature enough, it is impossible to accurately dismantle the bars in the scene with background noise and multi-person conversations, and there are usually several conference speakers in the meeting. heavy workload

[0006] For face recognition technology, nowadays, by calculating the distribution of the faces that appear in the video time and space, it is judged whether it is the host, but the conference is different from the news, and the splitting of the conference is to analyze the content of the speeches of the speakers in each conference. To break it down, because each conference speaker's speaking time is different, it is impossible to simply confirm the identity of the conference speaker based on the proportion of faces in the entire video

[0007] To sum up, the existing methods of dismantling news articles cannot be directly transferred to conference reports, and the method of dismantling articles used for conference reports today is usually manual dismantling, that is, manually previewing the conference video and performing dot dismantling. This method requires A large amount of manpower and low efficiency, so it is necessary to further improve the existing technology

Images