Video classification method, method, device and equipment for constructing classification model

Abstract

The invention discloses a video classification method, a method, a device and equipment for constructing a classification model, which can be applied to the field of video classification. The video classification method comprises the following steps: acquiring a to-be-classified video, and inputting the to-be-classified video into a classification model for video classification; wherein the classification model is obtained by training an initialization model by classified videos, and the initialization model is obtained by model parameters of a self-supervised model trained based on unclassified videos. In the process of obtaining the classification model, the model parameters applied to the classification model are obtained through training of the unclassified videos, the situation that a large number of classified videos are needed to train the classification model originally is avoided, the manual classification time is shortened, the training efficiency and precision of the classification model are improved, and the classification efficiency is improved. And thus, the video classification efficiency and precision are improved.

Description

technical field [0001] The present application relates to the technical field of video classification, and in particular, to a method for video classification, and a method, apparatus and device for constructing a classification model. Background technique [0002] At present, in the usage scenarios of video application software, there is a general need to classify videos. Generally, the method of manual review and annotation can be used to classify videos, and the staff will review the videos and perform category annotations. This process has the problems of low video classification efficiency and low video classification accuracy. [0003] With the continuous increase in the number of videos, it is difficult to meet business needs using existing classification methods, and improving the efficiency and accuracy of video classification has become the development direction to solve the problem of video classification. SUMMARY OF THE INVENTION [0004] The present applicati...

AI Technical Summary

Problems solved by technology

Usually, the method of manual review and labeling can be used to classify videos, and the staff will review the videos and perform category labeling. This process has the problems of low video classification efficiency and low video classification accuracy.

[0003] As the number of videos continues to increase, it is difficult to meet business needs using existing classification methods. Improving the efficiency and accuracy of video classification has become the development direction to solve video classification problems.

Images