Design of Video Advertisement Classification System under the Background of New Media

In order to obtain ideal mass video advertisement retrieval results, a video advertisement classification system under the background of new media is proposed. According to the feedback of the basic layer, the logic analysis and processing layer segments the original video advertisement sequence, uses multi tributary network to encode the intermediate frame sequence, extracts the useful information in the video advertisement, introduce the spatial attention prediction model to each tributary of the multi tributary network, integrates the attention spectrum the feature spectrum extracted from each tributary network. The total connection layer obtains the classification score of video advertisement, and linearly superimposes it to obtain the type estimation result of the overall video advertisement, and feeds tack the result to the user. The results show that the system can achieve video advertising classification, classification accuracy is high, and the overall performance of the system is superior. CCS CONCEPTS • Information systems∼Information retrieval∼Users and interactive retrieval