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

An image recognition method and device

An image recognition and image technology, applied in neural learning methods, character and pattern recognition, instruments, etc., can solve the problem of weak distinction of semantic knowledge and achieve the effect of improving recognition ability

Active Publication Date: 2022-07-29
UNIV OF SCI & TECH OF CHINA +1
View PDF10 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, due to the weak discriminative nature of semantic knowledge, it is difficult for this semantically aligned visual representation to separate the two domains, so images of unknown domains are more likely to be recognized as known domain categories.

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
  • An image recognition method and device
  • An image recognition method and device
  • An image recognition method and device

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0024] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to specific embodiments and accompanying drawings.

[0025] The embodiment of the present invention proposes a new domain-aware-based bias elimination technology to realize an unbiased zero-sample image recognition application. The core idea is to construct two complementary visual representations, namely, a semanticless visual representation and a semantic visual representation, to deal with samples in known and unknown domains, respectively. For semantic-free visual representations, an adaptive second-order embedding module can be designed to extract second-order statistics in visual information, and maximize its inter-class differences through an adaptive edge Softmax. This makes semantic-free visual representations sufficiently distinguishable to perform both category prediction for known...

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 present invention provides an image recognition method, comprising: S1, acquiring a training image set, training each training image in the training image set and a category index corresponding to the training image, and learning to extract non-semantic visual expressions; S2, extracting the training image Align each training image with the semantic label corresponding to the training image, and learn to extract semantically aligned visual expressions; S3, at the same time, identify and analyze non-semantic visual expressions and semantically aligned visual expressions to obtain a visual bias elimination model; S4, combine The image to be recognized is input into the visual bias elimination model for recognition. The image recognition method provided by the present invention can improve the perception effect of samples in the known domain and the unknown domain by establishing a visual bias elimination model, thereby realizing accurate recognition of zero samples without bias.

Description

technical field [0001] The invention relates to an application of realizing the recognition of unbiased zero-sample images based on a domain perception-based bias elimination technology, and in particular to an image recognition method and device. Background technique [0002] Zero-shot learning aims to simultaneously identify image samples of known class (known domain) or unknown class (unknown domain). Recent approaches focus on learning a semantically aligned visual representation to transfer knowledge from known domains to unknown domains. However, due to the weak discriminativeness of semantic knowledge, it is difficult for such semantically aligned visual representation to separate the two domains, thus, images of unknown domains tend to be recognized as known domain categories. SUMMARY OF THE INVENTION [0003] (1) Technical problems to be solved [0004] An image recognition method and device provided by the present invention are used to at least solve the above-...

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
Patent Type & Authority Patents(China)
IPC IPC(8): G06V20/70G06V10/774G06V10/778G06V10/82G06N3/04G06N3/08G06K9/62
CPCG06N3/08G06F18/214
Inventor 张勇东闵少波谢洪涛
Owner UNIV OF SCI & TECH OF CHINA
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

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com