Video data processing method and device

Abstract

The embodiment of the invention provides a video data processing method and device, and the method relates to the technical field of network media, and can achieve the processing of video data through employing a deep learning algorithm. The method comprises the following steps: clustering video tags in an obtained video tag set to obtain K tag clusters; obtaining a sample video category and a sample video tag corresponding to the sample video data, and determining a sample cluster class identifier of the sample video data according to a tag cluster to which the sample video tag belongs; outputting sample multi-modal features corresponding to the sample video data through the initial video multi-modal model; inputting the sample multi-modal features into N classification components, and outputting N classification results corresponding to the sample video data through the N classification components; and training the initial video multi-modal model according to the N classification results, the sample video category, the sample video tag and the sample cluster identifier. By adopting the embodiment of the invention, the effect of multi-modal representation learning of the video can be improved.

Description

technical field [0001] The present application relates to the technical field of network media, in particular to a video data processing method and device. Background technique [0002] In the context of Internet big data, it is usually necessary to process and analyze specific data and extract useful information from the data, that is, to perform representation learning on the data. How to use the massive data on the Internet to mine effective information has attracted widespread attention. At present, the representation learning of video data can adopt a supervised training method. By obtaining the supervision information of video data, the supervision information can be used to guide the classification of video data features; however, the existing supervision information is usually a single label marked by humans. , the video data features learned by using the supervision information are often coarse-grained features, and the predicted classification results of the video ...

AI Technical Summary

Problems solved by technology

At present, the representation learning of video data can adopt a supervised training method. By obtaining the supervision information of video data, the supervision information can be used to guide the classification of video data features; however, the existing supervision information is usually a single label marked by humans. , the video data features learned by using the supervision information are often coarse-grained features, and the predicted classification results of the video data features may be different from the content of the video data itself, which makes the effect of video data representation learning unsatisfactory.

Images