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

A Place Recognition Method Based on Knowledge Graph Reasoning

A technology of knowledge graph and recognition method, which is applied in the direction of reasoning method, knowledge expression, neural learning method, etc., can solve the problems of low recognition rate of recognition method, achieve good place recognition effect, simple steps, and easy to realize effect

Active Publication Date: 2022-05-13
SOUTHEAST UNIV
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] In order to overcome the deficiencies or deficiencies in the background technology, the present invention provides a place recognition method based on knowledge graph reasoning, which can effectively solve the problem of low recognition rate of the recognition method based on single information by combining knowledge graph technology to fuse various place environment information , and can enhance the semantic richness of reasoning results, which helps to improve human-computer interaction and other place-related intelligent tasks

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
  • A Place Recognition Method Based on Knowledge Graph Reasoning
  • A Place Recognition Method Based on Knowledge Graph Reasoning
  • A Place Recognition Method Based on Knowledge Graph Reasoning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0034] Embodiment 1: a kind of place recognition method based on knowledge map reasoning, comprises the following steps:

[0035] Step 1) basic semantic data acquisition,

[0036] The basic semantic data mainly describes the items contained in a specific place, the events that occur, and the special semantic concepts related to the place. All kinds of information, including images, sounds, distances, voices, etc., are described and marked in artificial natural language, so as to obtain basic semantic data and corresponding place categories. On the other hand, in the process of place recognition and reasoning, the above semantic information will be automatically generated according to the types of heterogeneous information, combined with the existing semantic generation model;

[0037] Step 2) place description entity generation,

[0038] Use natural language processing methods such as text segmentation, stop word removal, entity extraction, morphological restoration, and man...

specific Embodiment

[0053] Specific embodiments: the framework of a place recognition method based on knowledge map reasoning proposed by the present invention is as follows figure 1 As shown, it includes the training process and inference process respectively. Such as figure 1 As shown, the training process mainly includes the following four steps:

[0054] 1) Obtain basic semantic data by manual labeling from multiple types of heterogeneous place information, mainly the semantics of things covered by place information described in natural language, and use this as a data sample to design a semantic generation model;

[0055] 2) Use natural language processing methods to preprocess and screen basic semantic data to obtain the description entity knowledge of places;

[0056] 3) Through the sample statistics in the actual application environment, the occurrence probability corresponding to the description entity is obtained, thereby forming a place knowledge map with the basic triple structure o...

specific Embodiment 1

[0063] Specific embodiment 1 experimental process and result

[0064] The implementation process of the place recognition method based on knowledge map reasoning in the present invention will be further described below in conjunction with specific experiments and accompanying drawings. This embodiment is only a preferred example of the present invention and should not be construed as a limitation of the present invention.

[0065] The place information data used in the experiment of the present invention comes from J.Xiao et al. (SUN dataset.https: / / vision.cs.princeton.edu / projects / 2010 / SUN / ,2020-11-25. HaysJ, Ehinger K A, et al.SUN database:Large-scale scene recognition from abbey tozoo[C] / / Computer Vision&Pattern Recognition.IEEE,2010.) A large-scale scene image database, which contains 397 categories and a total of about 100,000 images RGB images, each scene contains at least 100 image samples, and about 16,000 images have been manually labeled, and the main objects contain...

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 place recognition method based on knowledge map reasoning, and proposes a general place recognition method based on knowledge map reasoning and capable of fusing various heterogeneous environmental information on the basis of the knowledge map construction method in the place field. , the steps are as follows: (1) Extract the main clues such as the main objects, events, and spatial structures that constitute the place from various heterogeneous information, and describe these clues in natural language; (2) Use natural language processing methods Screen the above descriptions to form place description entities; (3) combine the occurrence frequency of the above description entities in the actual environment to build a place domain knowledge map; (4) use the deep neural network to realize the reasoning classification based on the knowledge map, and give the final Recognition results; the present invention improves the accuracy of place recognition by using the knowledge map reasoning method, and greatly improves the semantic interpretability in the process of place recognition.

Description

technical field [0001] The invention relates to a location recognition method based on knowledge map reasoning, which belongs to the technical field of artificial intelligence and knowledge map. Background technique [0002] Place awareness refers to the use of environmental information such as vision, sound, distance, and natural language to automatically process and analyze it through artificial intelligence methods, and to judge and recognize the specific location semantics (such as kitchens, streets, etc.) . Place awareness not only helps to understand the overall semantic content of environmental information, but also provides a basis for human-computer interaction tasks related to places. role. [0003] Most of the existing place recognition technologies use images or distances (such as infrared rays, ultrasonic waves, etc.) as recognition clues, and learn and train the deep neural network model through massive samples, so that the network model can give the correspo...

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): G06F16/387
CPCG06N5/02G06N5/04G06F16/3344G06F16/3346G06F16/35G06F16/367G06F40/216G06F40/295G06F40/30G06N3/08G06N3/047G06N3/045G06F18/241G06F18/2415G06F40/268G06N5/022G06N3/042G06N3/044G06F40/44
Inventor 李新德李沛孙长银
Owner SOUTHEAST UNIV
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