A kind of picture style recognition method and device

Abstract

The present application provides a picture style recognition method and device. The method includes: obtaining sample pictures, processing the sample pictures according to a preset method to form a sample training set; performing parameter initialization on a preset multi-objective convolutional neural network and initializing the sample pictures in the sample training set in parameter initialization After training in the multi-objective convolutional neural network to obtain the picture style recognition model; use the picture style recognition model to identify the picture to be recognized, obtain the probability vector of the different style types of the picture to be recognized, and the probability value of each style type in the probability vector The value range of is from 0 to 1; identify the style type of the image to be identified according to the preset judgment rules and probability vector. Utilizing the technical solution provided by the embodiment of the present application, automatic and accurate recognition of the style of commodity pictures can be realized, the efficiency of picture style recognition can be improved, the work intensity of operators can be reduced, and user experience can be improved.

Description

technical field [0001] The present application belongs to the technical field of image information processing, and in particular relates to a method and device for recognizing a picture style. Background technique [0002] With the development of the Internet consumption era, consumers can choose their favorite products online, which greatly facilitates users to shop. For example, consumers can choose their favorite product types through the product pictures displayed by online merchants. [0003] Generally, when consumers purchase goods online, they are often influenced by various conceptual factors, such as brand, price, color, style, etc. These conceptual factors can generally be manually set by the merchant on the service operation platform. Among the many conceptual factors, some factors such as clothing brand, price, color, etc. are usually easy to define, and generally have relatively clear and standardized boundaries for distinction. For some other commodity concep...

AI Technical Summary

Problems solved by technology

However, both merchants and consumers will often have large deviations due to subjective factors when judging the style type of the product to be filled in. Therefore, the concept of product style type like this often causes confusion and ambiguity among merchants and consumers. Affect the style classification of merchant products or the selection of consumer product styles, reducing the effect of product marketing

This not only affects the transaction conversion rate, but also reduces the user experience

At the same time, due to the wide variety of online products, generally the amount of information involved in the processing of product styles is relatively large, which will also consume a lot of working time and labor of operators

[0004] In the prior art, the use of manual subjective judgment to identify product styles often leads to large differences in product style recognition results, making it difficult to perform accurate, reasonable and unified style recognition

At the same time, the judgment method of manual recognition also increases the work intensity of the operator's product style recognition and reduces the efficiency of style recognition

Images