Enhanced training method and device for image recognition model

An image recognition and training method technology, applied in the computer field, can solve the problems such as the inability to guarantee the security of funds in the face payment authentication business scenario, the reduction of the accuracy of the image recognition result, etc., to achieve the effect of good recognition function and improved robustness

Active Publication Date: 2022-08-09
ALIPAY (HANGZHOU) INFORMATION TECH CO LTD
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] In this case, if the image recognition model only uses standard training samples for training, the accuracy of the image recognition results may be reduced in the case of increased image interference, causing certain troubles, such as face payment authentication business scenarios. Fund security cannot be guaranteed

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
  • Enhanced training method and device for image recognition model
  • Enhanced training method and device for image recognition model
  • Enhanced training method and device for image recognition model

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0047] The solution provided in this specification will be described below with reference to the accompanying drawings.

[0048] First, combine figure 1 A specific implementation scenario is shown for illustration. like figure 1 As shown, it is a specific implementation scene of face recognition. In this implementation scenario, first, the image recognition model is trained through the first computing platform. The first computing platform can use a large number of face images as training samples to train an image recognition model. Wherein, the face image sample may correspond to at least one of the labeling result of the outline of a face, the labeling result of whether there is a living body, etc., and is used to guide the output result of the image recognition model. The image recognition model can be implemented by, for example, a convolutional neural network (CNN) or the like. The training process of the image recognition model can be carried out in any suitable con...

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 embodiment of this specification provides an enhanced training method for an image recognition model, wherein a predetermined number or a predetermined proportion of samples are randomly selected from a first sample set as seed samples, and expanded to obtain several expanded samples. The extended sample adds disturbance to the original image, while the labeling result remains unchanged. In the process of sample expansion, the disturbance value takes the reference pixel as a reference and presents a predetermined distribution around, so that the real disturbance can be well simulated. Since the labeling result of the extended sample remains unchanged after adding disturbance, the image recognition model trained by the extended sample can well identify the target recognition result of the original image, thereby improving the robustness of the image recognition model.

Description

technical field [0001] One or more embodiments of this specification relate to the field of computer technology, and in particular, to a method and apparatus for image disturbance processing, image sample expansion, and enhanced training of an image recognition model by using extended samples. Background technique [0002] Image recognition refers to the technology that uses computers to process, analyze and understand images to identify targets and objects in various patterns. Image recognition is widely used in various fields, such as unmanned driving, attendance, payment authentication, criminal investigation tracking and so on. Among them, in some business scenarios, in addition to target detection, it is also necessary to identify whether the relevant target is a real target entity. For example, in the face payment authentication scenario, considering the security of user funds, in order to avoid using face photos to pretend to be real people for payment , it is also n...

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): G06V40/16G06V40/40G06T5/00G06N3/04G06N3/08G06Q20/40G06V10/82
CPCG06T5/009G06Q20/40145G06N3/08G06T2207/10004G06T2207/20081G06T2207/20084G06T2207/20221G06T2207/30201G06V40/166G06V40/161G06V40/45G06N3/045G06N20/00G06V40/172G06N5/04G06F18/214G06F18/22G06F18/251
Inventor 徐文浩
Owner ALIPAY (HANGZHOU) INFORMATION 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