Clothing suit recommendation system and method

Abstract

The invention belongs to the technical field of intelligent clothing, and discloses a clothing suit recommendation system and method, and the system comprises an image acquisition unit, an image processing unit, a feature extraction unit, and a clothing recommendation unit. The image acquisition unit is used for acquiring suit images of enough clothes and related text descriptions of the suit images; the image processing unit is used for storing images, texts and feature vectors of the images and the texts in the suit-by-suit clothes in a database; the feature extraction unit is used for preprocessing collected suit images and texts of a large number of clothes, performing feature extraction and fusion on the suit images and texts one by one and generating matching scores of the clothes; and the clothing recommendation unit is used for performing similarity calculation on the extracted image features and features pre-stored in a database, and recommending the most similar clothing. The clothing suit and the description text of the clothing suit are processed by using the convolutional neural network and natural language processing, so that the overall speed of algorithm processing is greatly improved, and the timeliness of the system is improved.

Description

technical field [0001] The invention belongs to the technical field of smart clothing, and more specifically, relates to a novel clothing suit recommendation system and method. Background technique [0002] At present, when shopping online, it is usually difficult to start with a wide range of products. Customers generally have one or several suits that they like very much, but there are very limited scenarios where any combination of them will produce very good results. , People generally have the need to have jackets and trousers to match suitable scarves and hats, but the current method of similar matching through purchase records is not accurate and representative, and it is also very limited in matching existing clothing. [0003] The Chinese patent with the publication number CN111815351A "A Clothing Recommendation Method Based on Collaborative Filtering and Association Rules" collects clothing data, labels, purchase records and other information, calculates the histor...

AI Technical Summary

Problems solved by technology

[0003] The Chinese patent with the publication number CN111815351A "A Clothing Recommendation Method Based on Collaborative Filtering and Association Rules" collects clothing data, labels, purchase records and other information, calculates the historical shopping preferences of members within the target time through a collaborative filtering algorithm, and trains Corresponding shopping probability prediction model, so as to give a recommendation ranking result of shopping probability; this method has the problem that it can only be recommended based on purchase records, and the data used for calculation is very limited, and it is designed with a small amount of data Probabilistic model, so the recommendation accuracy is not high

The Chinese patent "A Method and Apparatus for Recommending Clothing Products" with application number 2CN201210254720.1 queries the pre-established correspondence between the user ID and the attribute value of the first attribute of the clothing product according to the user ID, and obtains the operation The attribute value corresponding to the user ID carried in the request, but it uses attribute keywords to recommend clothing, which is prone to interference from irrelevant keywords, and the algorithm is relatively complicated

Images