Multi-source heterogeneous data identity recognition method based on attention mechanism

A multi-source heterogeneous data and identity recognition technology, which is applied in the field of multi-source heterogeneous data identity recognition based on the attention mechanism, can solve problems such as the inability to obtain accurate recognition results, and overcome the error recognition of a single face camera to capture faces picture effect

Active Publication Date: 2019-07-16
CHINACCS INFORMATION IND
View PDF3 Cites 9 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

There is a problem in these methods, that is, the interference information of the original data is required to be as little as possible, otherwise accurate recognition results cannot be obtained.

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
  • Multi-source heterogeneous data identity recognition method based on attention mechanism
  • Multi-source heterogeneous data identity recognition method based on attention mechanism
  • Multi-source heterogeneous data identity recognition method based on attention mechanism

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0037] see Figure 1 to Figure 4 , the present invention provides a multi-source heterogeneous data identification method based on an attention mechanism, which specifically includes the following steps:

[0038] Step S1: install multiple face cameras on the pedestrian route, capture face pictures, and record the time and location of the capture;

[0039] Step S2: Use the built face recognition system to identify face pictures, and compare the face pictures in the blacklist database to return the numbers in the 10 blacklist databases with the highest similarity to the face pictures in the blacklist database ID, and comparison similarity result data;

[0040] Step S3: On the pedestrian trajectory, the time interval ΔAB<ΔT means that the time interval between the two groups of IDs does not exceed the maximum time threshold and the conditional rule of the spatial trajectory. Compare the IDs returned by the face pictures captured at different capture locations, Combine IDs with ...

Embodiment 2

[0059] This solution includes a method of selecting the best combination based on the ID number IDs, specifically:

[0060] First, use face recognition technology to identify the face pictures captured at multiple points and return the face picture ID numbers in the blacklist library and the similarity Sim of each ID;

[0061] Compare the IDs returned by each point to select the best combination of multi-point IDs;

[0062] Encode the external data such as the time cut-off point of the face picture capture and the weather into an embedding vector v1 through one-hot;

[0063] The IDs similarity Sim and the embedding vector are serially input into the local attention mechanism unit, and the similarity of each identity number ID is reassigned;

[0064] The weather data, time cut-off point, and surrounding building data of multiple face camera capture points are encoded into an embedding vector v2 by one-hot, and the embedding vector v2 is input into the spatial attention mechani...

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 multi-source heterogeneous data identity recognition method based on an attention mechanism, relates to the field of equipment, and adopts the technical scheme that a plurality of face cameras are mounted on a pedestrian advancing route, face images are snapshot, and the snapshot time is recorded at the same time; the face images are recognized through the built face recognition system, and similarity data is returened; an IDs similarity vector is obtained; offline training is carried out on the model through a cross entropy loss function by utilizing a large amount of label data; and identity recognition is realized. The method has the advantages that the face images are captured through the face camera capture points, the problem that the recognition accuracy isnot high due to the fact that the method is limited by a non-matching scene of capturing the face images through a single camera is solved, and the method is more friendly and more rapid and effective; external influence factors of a human face picture snapshot scene are simulated by utilizing external weather data, time intercept points and surrounding building data, and the human face similarity of a real scene can be restored as much as possible.

Description

technical field [0001] The invention relates to the technical field of identification, in particular to an attention mechanism-based multi-source heterogeneous data identification method. Background technique [0002] At present, identification methods are becoming more and more mature, and most of them focus on learning and processing biological characteristics (such as face, fingerprint, iris, voiceprint) to judge the identity of the identification object; or to identify the dynamic characteristics of the object (such as walking posture, movement, Trajectory), by learning to compare the similarity, so as to judge the identity of the recognition object. There is a problem in these methods, that is, the interference information of the original data is required to be as little as possible, otherwise accurate recognition results cannot be obtained. Therefore, finding an efficient and practical identification method will have a wide range of practical applications in the field...

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 Applications(China)
IPC IPC(8): G06K9/00
CPCG06V40/172G06V40/161Y02A90/10
Inventor 舒泓新蔡晓东黄玳王秀英
Owner CHINACCS INFORMATION IND
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap