Method for embedding artificial intelligence feature information based on AVS3 coding framework

Abstract

The invention discloses a method for embedding artificial intelligence feature information based on a third-generation audio and video standard (AVS3) coding framework. The method comprises the following steps: S1, acquiring an original image which is not coded through a camera; S2, performing feature extraction on the original image by using an artificial intelligence algorithm engine, performing AVS3 coding on the original image, and outputting a coded AVS3 code stream; S3, after AVS3 image coding, newly establishing an artificial intelligence coding extended data structure in an extended data area in an AVS3 coding framework, and filling the features and related data information into the artificial intelligence coding extended data structure; and S4, after the features and the related data information are filled into the artificial intelligence coding extended data structure, outputting an AVS3 code stream with the features. According to the method, the problem that no artificial intelligence feature is carried in an AVS3 coding framework is solved, the AVS3 coding standard is further perfected, and meanwhile, an efficiency balancing mechanism can be provided for the cloud end and the side end in market application.

Description

technical field [0001] The invention belongs to the field of video compression coding, and in particular relates to a method for embedding artificial intelligence feature information based on the coding framework of the third-generation audio and video standard (AVS3). Background technique [0002] With the continuous upgrading and progress of artificial intelligence technology, a large number of products based on end-side artificial intelligence target feature detection and capable of high-definition or ultra-high-definition video encoding have appeared on the market. In the current common environment, the preliminary target feature detection is generally carried out on the terminal side, and the target features are uploaded to the cloud for subsequent subsequent processing, such as in-depth data analysis, which can alleviate the multi-channel acquisition of the cloud to a certain extent. Computing power requirements for video processing, which can also bring lower hardware...

AI Technical Summary

Problems solved by technology

[0003] Based on this, the present invention provides a method of embedding artificial intelligence feature information based on the AVS3 coding framework, which solves the problem of no artificial intelligence feature carrying in the AVS3 coding framework, further improves the AVS3 coding standard, and can also be applied in the market Provides an efficiency balancing mechanism for the cloud and the edge

Images