Method for recommending similar pictures based on picture set

Abstract

The invention discloses a method for recommending similar pictures based on a picture set, and the method comprises the steps: S1, extracting a plurality of picture features of the picture set througha picture extraction model, and carrying out the clustering of pictures in the picture set, so as to extract at most two pictures with the most picture features, extracting visually similar picturesfrom a target recommendation pool according to the at most two pictures; and S2, extracting title vectors from the picture set through an algorithm, and extracting pictures with similar texts in a target recommendation pool according to the title vectors. According to the method for recommending the similar pictures based on the picture set, the most representative picture is found in the pictureset through the clustering method, then the visual similar picture is found through the most representative picture, and the visual similar picture is combined with the text similar picture, so that the content of the related recommended pictures is enriched, and too large calculation amount of a visual similarity algorithm module is avoided, and feasibility of system implementation is considered.

Description

technical field [0001] The invention belongs to the technical field of similar picture recommendation methods, and in particular relates to a method for recommending similar pictures based on a picture set. Background technique [0002] Picture recommendation, also known as picture-related recommendation, refers to recommending relevant content to the user based on the content currently viewed by the user. For example, if the picture currently viewed by the user is St. Suggest pictures of churches or other buildings, etc. Picture recommendation is not only conducive to the promotion of picture information, but also can provide users with more potential resources that may be of interest, which is convenient for users to further search or obtain. Therefore, in the existing information promotion and other fields, how to optimize picture The recommendation is of great significance for improving the overall performance of the information promotion scheme. [0003] However, reco...

AI Technical Summary

Problems solved by technology

[0003] However, recommending similar pictures through pictures often has certain limitations. You can only recommend similar pictures from a single picture, which often cannot meet the needs of users. For example, users want to find similar pictures through the picture collection for incremental supplementation. , so there is a method of recommending a picture set through a picture set

[0004] However, usually the content recommended by the recommendation algorithm based on picture collections is picture collections. From the perspective of end users, what users want to be recommended is the pictures themselves; recommended picture collections are also difficult to meet the needs of finding relevant pictures to supplement the content of existing picture collections, for example When you are interested in a certain picture collection, you want to get more pictures, etc.

[0005] Therefore, there is a method that can recommend pictures in picture collections, but the traditional algorithm for recommending pictures based on picture collections is often based on the similarity of text content, but many picture collections may not have information such as text descriptions at all.

Images