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

Picture classification method based on deep transfer learning

A technology of transfer learning and image classification, applied in the field of artificial intelligence and big data processing, which can solve problems such as unlabeled classification

Active Publication Date: 2019-03-26
深圳市快麦科技有限公司
View PDF7 Cites 25 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

An image classification method based on deep transfer learning is proposed to solve the unlabeled classification problem in image classification

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
  • Picture classification method based on deep transfer learning
  • Picture classification method based on deep transfer learning
  • Picture classification method based on deep transfer learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0030] The technical solutions in the embodiments of the present invention will be described clearly and in detail below with reference to the drawings in the embodiments of the present invention. The described embodiments are only some of the embodiments of the invention.

[0031] The technical scheme that the present invention solves the problems of the technologies described above is:

[0032] figure 1 A flow chart of a method for classifying images based on deep transfer learning provided by the present invention, specifically comprising:

[0033] Data preparation stage. Determine the image sample set of clothing products to be classified as the target domain sample data set D target , Determine the category set C to be divided into, the main category contained in it is , download the ImageNet data set, and select the image sample set containing the Clothing mark as the source of the existing mark domain sample dataset D source , and D source According to the category...

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 requests to protect a picture classification method based on deep transfer learning. wherein the domain adaptation at least comprises data of two domains; wherein the domains are respectively a source domain and a target domain; wherein the source domain data is marked sample data, and the source domain data is marked sample data. The method comprises the following steps: 1) a data preparation stage: preparing source domain data and target domain data, The method comprises the steps of (1) establishing a target category set, (2) establishing a feature extraction model, (3) establishing a basic feature extraction model by using ResNet and a self-attention network, (4) establishing a domain confrontation model, and (5) predicting a sample category and a sample domain by using the domain confrontation model. Wherein domain marking is conducted on a source domain sample and a target domain sample, and a loss function based on the sample migration weight is set, and 5, prediction is conducted on target domain data, a category prediction result serves as a final result, the marking cost is reduced, and the purpose of knowledge migration is achieved.

Description

technical field [0001] The invention belongs to the technical field of computer information processing, and specifically relates to the related fields of artificial intelligence and big data processing. Background technique [0002] With the increasing data scale and computing resources, big data processing technology is also experiencing rapid development. As one of the effective tools of big data processing technology, machine learning plays a key role in big data processing technology. Supervised learning is the An important branch of learning, which is characterized by learning that includes label information. In real life, labels for certain tasks are often difficult to obtain, such as image data, which requires a lot of manpower to label. The largest existing label The image data set is ImageNet, which contains 15 million labeled image data. Its labeling task was completed by 48,940 workers from 167 countries in 2 years. Therefore, how to reduce the labeling cost is a ...

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): G06N3/04G06K9/62
CPCG06N3/045G06F18/00
Inventor 王进王科李林洁杨俏孙开伟刘彬
Owner 深圳市快麦科技有限公司
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