Garment attribute editable garment image retrieval method

Abstract

An embodiment of the invention discloses a garment attribute editable garment image retrieval method. The method comprises the steps of extracting eigenvectors of garment images in an existing garmentdatabase; obtaining a garment image used to be searched for and input by a user; extracting an eigenvector of the garment image used to be searched for by the user; obtaining garment attributes of the garment images by using a garment attribute judgment method; displaying the garment attributes and selectable modification items to the user, and obtaining garment attribute items modified by the user; by using the eigenvector of the originally searched garment image and the garment attribute items modified by the user, forming a new garment eigenvector; calculating the similarity between a newly formed garment attribute feature and garment attribute features in the garment database; and displaying the first k garment images with the highest similarity. By implementing the method, the user can modify the attributes of the to-be-retrieved garment image according to own preferences and then perform image content-based garment retrieval to obtain a satisfied query result.

Description

technical field [0001] The invention relates to the field of image technology, in particular to a clothing image retrieval method with editable clothing attributes. Background technique [0002] With the rapid popularization of the Internet and the rapid development of e-commerce, people's online shopping behavior is more common and frequent, and there are more and more categories of goods. Online product retrieval systems have become indispensable for users to purchase goods online. tool. At the same time, among many commodities, clothing is one of the commodities that people often buy. Therefore, an efficient and functional clothing retrieval system is of great significance to online shopping platforms, especially, it is an indispensable function for e-commerce platforms that specialize in online shopping for clothing. Traditional clothing image retrieval methods generally label clothing images manually first, and then use keywords to retrieve corresponding clothing imag...

AI Technical Summary

Problems solved by technology

The method of clothing attribute feature generation is used to generate clothing attribute features, and then fused with the original clothing image features to form new clothing features that meet the user's intentions, and then combined with image content-based clothing retrieval methods to solve the original image content-based problem. The clothing retrieval method can only retrieve according to the image content queried by the user, and the user cannot edit the relevant attributes of the clothing image used for query according to his own preferences, so that the user can modify the attributes of the clothing image to be retrieved according to his own preferences, and obtain satisfactory results. search result

Images