A method, device and system for structuring event video

Abstract

The invention discloses a method, device and system for structuring event video. The method at least includes the following steps: obtaining N video frames of the event video stream to be processed, identifying the event time identifier in the video frame, and performing the first step with the display time stamp. A time axis match, and then analyze and obtain the event and perform a second time axis match on the occurrence time of the event in the event and the display timestamp, and determine the starting point and display timestamp and end point of the target event based on several associated events that constitute the target time Display the timestamp, and finally extract the video frame of the target event and edit and compress it into the target video. This method can use two time axis alignments and deep learning algorithms to realize automatic editing of batch event videos, effectively improving the timeliness and accuracy of video editing. Realize that there is no need to pull all the video streams to be processed, and only extract a few relevant video frames for processing, reducing network bandwidth costs and server resource consumption, and reducing costs.

Description

technical field [0001] The present invention relates to the field of video processing, in particular to a method, device and system for structuring event video. Background technique [0002] In the operation of copyrighted events, traditional media will edit some event clips for important events in the process of live broadcast of events or historical video on demand, such as goals, fouls and other events in football matches. The clipped event clips are used for users to browse and share. . Existing practices often rely on hand-editing production by a large number of editors. The timeliness is poor. As far as live events are concerned, the video of an event is often delayed by several minutes or even ten minutes before the event occurs, which affects the user's viewing experience. The content output is low, limited by operating resources, and content production can only be Focus on ensuring key events, the content output of non-head events is particularly affected, and the...

AI Technical Summary

Problems solved by technology

[0004] It can be seen that although this solution is intelligent, in the production process, editors still need to use the selection and editing module to select event segments based on statistical analysis results and feature labels, and drag and drop the time segments corresponding to event segments to the timeline to complete Event editing has not achieved completely unmanned operation, and the entire event data needs to be obtained during the processing process, the cost of network broadband and server resource consumption are high

Images