An image-based commodity recommendation method and system

Abstract

An embodiment of the invention discloses a commodity recommendation method based on an image and a system thereof and belongs to the Internet field. The method and the system can be suitable for a common multi-target scene and a good recommendation effect can be acquired. The method comprises the following steps of acquiring an image and carrying out pre-processing on the image; identifying a target area in the preprocessed image; aggregating the identified target area so as to generate a multimodal content description; according to the generated content description, acquiring a preliminary commodity recommendation result; and combining the acquired preliminary commodity recommendation result, deleting a repeated item, and outputting the combined commodity recommendation result.

Description

technical field [0001] The invention relates to the Internet field, in particular to an image-based product recommendation method and system in e-commerce. Background technique [0002] The recommendation system of the e-commerce platform can automatically display products that users may be interested in based on the user's past data and other relevant information, which greatly improves the user's purchasing efficiency and enhances the user experience. It has become an important part of the e-commerce platform. The product recommendation method based on image media data, as an emerging recommendation method, can mine and provide corresponding recommendation objects according to the media visual content that users are interested in, without user text input. As media content analysis technology advances, so does its precision. [0003] At present, image-based product recommendation methods generally assume that the product is the main content in the image, and return corresp...

AI Technical Summary

Problems solved by technology

However, this also determines that these methods are only suitable for commodity images of a single subject, and have stricter requirements on images and users, and are not suitable for actual occasions of casual shooting or scanning

At the same time, its recommendation results are completely dependent on the identification or matching of a single commodity entity, and the existing technology still does not have a high accuracy rate for the recognition of a large range of commodity images.

Images