Target article identification method and device, equipment, medium and product

Abstract

The invention discloses a target article identification method and device, equipment, a medium and a product. The method comprises the steps of obtaining an article picture which is to be identified whether the article picture contains a target article or not; performing multi-scale coding and decoding on the article picture to obtain a plurality of pieces of image feature information of captured article contour features, and extracting an article segmentation image from the article picture according to the plurality of pieces of image feature information; classifying each piece of image feature information to obtain a classification probability that the image feature information contains the target article, and solving an average probability of a plurality of classification probabilities obtained by classifying all the image feature information; performing image recognition on the article segmentation image to obtain a recognition probability that the article segmentation image contains the target article; and fusing the average probability and the recognition probability to obtain a fusion probability for result judgment, and when the fusion probability is greater than a preset threshold value, judging that the article picture contains the target article. According to the invention, the target article can be identified from the article picture more accurately.

Description

technical field [0001] The present application relates to the technical field of image recognition, and in particular to a target item recognition method and its corresponding device, computer equipment, computer-readable storage medium, and computer program product. Background technique [0002] The use of artificial neural network models for image recognition has become the mainstream of technology. For application scenarios such as e-commerce platforms, a large number of product pictures or other pictures are generated every day. For various application needs, including automatic product classification, product compliance detection, target product search, etc., it is generally necessary to Detect and identify the target items in the product pictures uploaded by users, and then do further processing according to different target items and needs. [0003] For application scenarios such as e-commerce platforms, if the product pictures uploaded by merchants to display items ...

AI Technical Summary

Problems solved by technology

[0003] For application scenarios such as e-commerce platforms, if the product pictures uploaded by merchants to display items include non-sale items, such as knives and swords, due to the characteristics of such products as small, reflective, and diverse forms, existing Sometimes it is difficult for the neural network model to accurately identify it from the product picture, which will cause great troubles for the e-commerce platform

It is unrealistic to rely entirely on manual investigation

Images