Method and device for classifying videos

Abstract

A video classification method and device, which relate to the field of electronic information technology and can improve the accuracy of video classification. The method comprises: conducting segmentation on a video in a sample video library in accordance with a time sequence, so as to obtain a segmentation result and generate a motion atom set; by using the motion atom set and the segmentation result, generating a motion phrase set which can express a complex motion mode, and generating a description vector of the video in the sample video library based on the motion phrase set; and by using the description vector, determining a video to be detected which is of the same type as the video in the sample video library. The method is suitable for a video classification scenario.

Description

technical field [0001] The invention relates to the technical field of electronic information, in particular to a video classification method and device. Background technique [0002] With the massive increase of video data, users browse videos one by one, and it takes a lot of time and energy to classify videos according to the motion information of characters in the videos. Although it is possible to classify videos for some simple sports such as walking and running in the video, the sports in the video are often more complicated, such as sports activities, etc., and video classification for simple sports can no longer meet the needs of users. In order to be able to classify videos for the more complex and continuous motion in these videos, the existing technology extracts some local area features from the videos, such as HOG (Histogram of Oriented Gradients, histogram of oriented gradients) and other features, based on these features. Clustering forms moving atoms, which...

AI Technical Summary

Problems solved by technology

[0004] However, for relatively continuous and long-lasting complex motions, it is difficult for the prior art to properly decompose such complex motions into segments composed of simple motions, and when the decomposition time points of the complex motions in the video are set differently, the same as The comparison scores obtained after comparing the fragments decomposed from the sample are also different, resulting in multiple different results for video classification, which are difficult to unify, and the accuracy of video classification is relatively low

Images