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

Image Pixel Semantic Annotation Method Fused with Multi-granularity Context Information

A technology of semantic annotation and image pixels, which is applied in the direction of instruments, character and pattern recognition, computer parts, etc., can solve the problems of identification and confusion, achieve high image annotation accuracy, realize the effect of intelligence, and reasonable design

Active Publication Date: 2018-06-05
TAIYUAN UNIV OF TECH
View PDF2 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0007] The purpose of the present invention is to provide an image annotation method using multi-granularity context information to solve the problem that the image annotation model has the ability to identify confusing features on the basis of satisfying the continuity of local annotations

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
  • Image Pixel Semantic Annotation Method Fused with Multi-granularity Context Information
  • Image Pixel Semantic Annotation Method Fused with Multi-granularity Context Information
  • Image Pixel Semantic Annotation Method Fused with Multi-granularity Context Information

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0032] Specific embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings.

[0033] In the process of image understanding, context information plays an important role in different types of target recognition. The traditional second-order conditional random field model only describes local context information. In order to obtain global context information, the present invention uses the neighborhood base The clique is expanded into context windows of two grain sizes, and the continuity of semantic label categories at adjacent positions is considered in the fine-grained context window, and the homogeneous transfer probability of semantic labels is calculated by using label smoothing parameters and fine-grained context descriptors. In the granular context window, sparse representation is used to describe the co-occurrence of different types of semantic tags, and the spatial co-occurrence relationship of semantic categories...

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 relates to the field of image intelligent recognition, in particular to a method for performing pixel semantic labeling on images by fusing multi-granularity context information. The present invention analyzes and counts the relationship between image labeling results and visual features, and forms a fine-grained position pair labeling model through fine-grained context description and labeling smoothing parameters to represent the transmission characteristics of semantic labels in a local area. At the same time, Coarse-grained contextual descriptors and semantic co-occurrence parameters are used to form a coarse-grained position pair labeling model to describe the co-occurrence relationship of semantic categories contained in images. Combined, the labeling model incorporates rich image information and has high image labeling accuracy, and then uses the piecewise method combined with the training data to perform segmented parallel training on the model parameters, which improves the training efficiency.

Description

technical field [0001] The invention relates to the field of image intelligent recognition, in particular to a method for performing pixel semantic labeling on images by fusing multi-granularity context information. Background technique [0002] With the advancement of image processing and analysis technology, the gradual improvement of computer performance and the continuous growth of the number of images, how to make computers see and understand the world like humans has become an important research goal of computer vision. It is one of the challenges in the field of artificial intelligence and computer vision to let the computer automatically interpret the content of the picture through computer programming to achieve image understanding. [0003] The research content and technical route of image understanding are currently mainly divided into three methods. One is to semantically label the image as a whole, using a certain amount of labels to provide high-level semantic ...

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): G06K9/62
CPCG06F18/2415G06F18/241
Inventor 谢刚赵婕赵文晶续欣莹杨云云
Owner TAIYUAN 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

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