A feature retrieval method and device, storage medium and computer equipment

A technology of feature collection and original feature, applied in the field of information service, can solve the problems of reduced retrieval accuracy, large amount of information, slow processing speed, etc., and achieve the effect of improving retrieval accuracy and speed.

Active Publication Date: 2022-06-03
SHENZHEN SENSETIME TECH CO LTD
View PDF11 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Therefore, when performing feature retrieval, the input features to be retrieved are searched among 1.4 billion known features, and the feature itself contains a relatively large amount of information, resulting in a very slow processing speed
[0003] In related technologies, by matching the known compressed features after the known features are compressed with the compressed features corresponding to the features to be retrieved, the known features corresponding to the matched known compressed features are used as the final retrieval result. In this way, by compressing Feature retrieval improves retrieval efficiency, but greatly reduces retrieval accuracy

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 feature retrieval method and device, storage medium and computer equipment
  • A feature retrieval method and device, storage medium and computer equipment
  • A feature retrieval method and device, storage medium and computer equipment

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0116] This embodiment provides a feature retrieval method, such as figure 2 As shown, the method includes the following steps:

[0117] S201. Perform feature extraction on the feature to be retrieved to obtain the compressed feature to be retrieved;

[0118] After receiving the retrieval operation, the retrieval device responds to the retrieval operation and generates features to be retrieved according to the retrieval images corresponding to the retrieval operations, wherein the features to be retrieved are the feature information of the retrieval objects included in the retrieval images, and the retrieval objects may be human faces, vehicles etc. objects. The retrieval device generates a retrieval request according to the retrieval feature, and sends the generated retrieval request to the service node.

[0119] After receiving the retrieval request, the service node parses the retrieval request to obtain the features to be retrieved carried by the retrieval request, and ...

Embodiment 2

[0173] In this embodiment of the present invention, by image 3 The shown network structure further describes the feature retrieval method provided by the embodiment of the present invention. image 3 The shown network structure includes: an interface proxy service (shard-proxy) 301, a service node 302, a database 303 and an object storage 304; wherein, the service node 302 includes a worker process (worker) and GPU / CPU, and the memory of the service node stores Compression features used by workers for retrieval. Each replica set (ReplicaSet) includes two service nodes: the master service node and the slave service node, where the worker of the master service node is the master process (master), the worker of the slave service node is the slave process (slave), and the master service node The first subset is stored in the video memory of the slave service node, and the second subset is stored in the video memory of the slave service node. The first subset of master service n...

Embodiment 3

[0256] In the embodiment of the present invention, four retrieval methods are used to compare the feature retrieval method in the related art with the feature retrieval method provided by the embodiment of the present invention, wherein the method 1 and the method 2 are the feature retrieval methods in the related art, and the method 3 And method 4 is the feature retrieval method provided by the embodiment of the present invention.

[0257] Method 1. Original Feature Retrieval

[0258] Figure 7A is a schematic diagram of the feature retrieval method 1 in the related art, such as Figure 7A As shown, the original feature array in the database includes a plurality of original features. When the feature to be retrieved is retrieved, the feature to be retrieved is matched with each original feature in the original feature array, and target candidates matching the feature to be retrieved are found. feature.

[0259] Method 2. Compressed Feature Retrieval

[0260] Figure 7B i...

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

Embodiments of the present invention provide a feature retrieval method and device, storage medium, and computer equipment, wherein the method includes: performing feature extraction on the feature to be retrieved to obtain the compressed feature to be retrieved; The target compressed feature set matched by the compressed feature, the target compressed feature set includes at least one target compressed feature, and the copy set includes different compressed features; determine the candidate feature corresponding to each target compressed feature from the original feature set, A candidate feature set is formed; the original feature set includes at least one original feature; the candidate features in the candidate feature set are compared with the features to be retrieved to obtain target candidate features corresponding to the features to be retrieved.

Description

technical field [0001] The invention relates to the field of information services, in particular to a feature retrieval method and device, a storage medium and computer equipment. Background technique [0002] The feature retrieval service is to find a feature matching the input feature to be retrieved from a series of known features. A series of existing known features are stored in the database, but feature-based retrieval services are usually used in intelligent video analysis, security monitoring and other fields. The known features stored in the database are massive, such as: national citizen face information database The facial features stored in the database are the facial features of 1.4 billion citizens across the country, including as many as 1.4 billion known features. Therefore, during feature retrieval, the input feature to be retrieved is searched among 1.4 billion known features, and the feature itself contains a relatively large amount of information, result...

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): G06F16/58G06F16/78
Inventor 陈宇恒樊俊良
Owner SHENZHEN SENSETIME TECH CO LTD
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