Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

GAN-based data enhancement unsupervised trademark retrieval system and method

A retrieval system and unsupervised technology, applied in the field of artificial intelligence, can solve problems such as difficult data labeling and insufficient collection

Inactive Publication Date: 2020-06-12
GUANGDONG UNIV OF TECH
View PDF4 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] In order to solve the problem of difficult data labeling and insufficient collection in the prior art, the present invention provides a GAN-based data-enhanced unsupervised trademark retrieval method

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • GAN-based data enhancement unsupervised trademark retrieval system and method
  • GAN-based data enhancement unsupervised trademark retrieval system and method
  • GAN-based data enhancement unsupervised trademark retrieval system and method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0041] An unsupervised trademark retrieval system based on GAN data enhancement, including a GAN data enhancement module, an instance distinction module, and a trademark retrieval module, wherein the GAN data enhancement module is used to enhance the trademark data set and expand the trademark training set; the instance distinction training module uses To train the unsupervised network to obtain the trademark feature extractor; the trademark retrieval module is used to calculate the similarity measure between the trademark database and the trademark features to be retrieved, and output the ranking results of the trademarks.

[0042]In a preferred solution, the GAN data enhancement module generates a trademark data set through a trained GAN model, and combines the enhanced trademark data set with the original trademark data set to form a new trademark database M .

[0043] In a preferred solution, the training steps of the GAN model are as follows:

[0044] Step 1. Keeping the...

Embodiment 2

[0060] A GAN-based data-enhanced unsupervised trademark retrieval method, applied to the above-mentioned system, includes the following steps:

[0061] S1. The GAN data enhancement module is used to enhance the trademark data set, expand the trademark training set, and obtain a new trademark database;

[0062] S2. The example distinguishing training module trains the unsupervised network, obtains the trademark feature extractor, and extracts the image features of the trademark to be retrieved;

[0063] S3. The trademark retrieval module calculates the similarity measure between the trademark database and the characteristics of the trademark to be retrieved, and outputs the ranking result of the trademark.

[0064] In a preferred solution, the GAN data enhancement module generates a trademark data set through a trained GAN model, and combines the enhanced trademark data set with the original trademark data set to form a new trademark database M .

[0065] In a preferred solut...

Embodiment 3

[0082] The invention provides a trademark retrieval method. Using ResNet50 as an unsupervised network, in the instance discrimination mode, the trademark data set is generated by the trained GAN model to generate a data set Q, and the data set Q is added to the original trademark data set to form a new trademark data set M, and finally passed The new trademark data set M is used to train the instance discrimination module to obtain the trademark feature extractor ResNet50. In the retrieval module, the new trademark data set M is extracted through the trained trademark feature extractor ResNet50 to form a trademark feature library F={F 1 , F 2 ,...,F n}. Similarly, the image to be retrieved is passed through the trained trademark feature extractor ResNet50 to extract the feature F', and finally, the similarity measure between the trademark database and the retrieved image is calculated according to the Euclidean distance, and sorted according to the size of the similarity me...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention discloses a GAN-based data enhancement unsupervised trademark retrieval system. The unsupervised trademark retrieval system comprises a GAN data enhancement module, an instance distinguishing module and a trademark retrieval module. The GAN data enhancement module is used for generating a trademark data set and expanding a training set; the instance distinguishing module is used fortraining an unsupervised network and extracting trademark features; and the trademark retrieval module is used for calculating the similarity between the trademark database and the trademark featuresto be retrieved and sorting the trademark features according to sizes. First, a trademark data set is used to train an adversarial generative network. Then, the trained GAN module is used for generating an enhanced data set, and an original trademark data set is added to form a new trademark data set; and finally, applying the new trademark data set to a training instance distinguishing module. Inthe trademark retrieval module, trademark features of a trademark image to be retrieved and a new trademark data set are extracted through a trained instance distinguishing module. According to the method, the problems of difficult data labeling and insufficient data diversity in trademark retrieval are effectively solved.

Description

technical field [0001] The invention relates to the field of artificial intelligence, in particular to a data-enhanced unsupervised trademark retrieval system method based on an adversarial generation network. Background technique [0002] As an important part of intellectual property rights, trademark protection has a profound impact on corporate brand value. At the same time, trademarks symbolize the quality of goods and the reputation of merchants to a certain extent. With the rapid development of the commodity economy, whether it is the number of trademark registration applications, trademark registration examinations or valid registered trademarks, it is almost impossible to manually find similar trademark images. Therefore, the trademark search system is considered to be a powerful tool for law enforcement agencies to handle trademark protection cases. However, at present, it takes a long time for trademark applicants to apply for a new trademark to approval, and mos...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06F16/583G06K9/62G06N3/04G06N3/08G06Q50/18
CPCG06F16/583G06N3/08G06Q50/184G06N3/045G06F18/22
Inventor 梁观术曹江中戴青云黄云飞
Owner GUANGDONG UNIV OF TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products