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

A multi-label classification method for image data

A technology of image data and classification method, which is applied in the field of neural network, can solve the problem that the secondary correlation and multiple correlation of labels cannot be extracted at the same time, so as to reduce the difficulty of prediction and improve the accuracy

Active Publication Date: 2021-04-20
ZHEJIANG UNIV
View PDF8 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, the current method cannot extract the secondary correlation and multiple correlation of tags at the same time.

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 multi-label classification method for image data
  • A multi-label classification method for image data
  • A multi-label classification method for image data

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0054] The present invention will be further described below in conjunction with drawings and embodiments.

[0055] Embodiments of the present invention are as follows:

[0056] The label set of the specific implementation data set is {plant, bird, person, sky, cloud, grass, mountain}, figure 1 is a sample of this dataset. The input image x of this sample is the pixel point of the image. It can be seen that the label of the sample is {bird, sky, cloud}, and the label of the image is y=[0, 1, 0, 1, 1, 0, 0].

[0057] figure 2 A model structure of the invention is shown.

[0058] step one:

[0059] Obtain the image vector x=[x from the input image 1 , x 2 ,...,x p ] T Get the image feature set V emb ={v 1 x 1 , v 2 x 2 ,...,v p x p}, where x 1 , x 2 ,...,x p respectively represent the gray value of each pixel in the input image, x∈R p , R p Represents the p-dimensional real number space as the input space, p represents the total number of pixels in the inpu...

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-label classification method for image data. Decompose the input image, use the neural network to extract the high-order correlation of the features, decompose the label data, use the neural network to extract the high-order correlation of the label, and use the neural network containing multiple fully connected layers to convert the feature code of the input image Decode from the input space to the label space; construct a loss function, initialize the training parameters, use the stochastic gradient descent method to minimize the final loss function as the goal, and train and solve to obtain the optimal training parameters; then input the image data to be tested into the trained The prediction is made in the model, and the label result is output to realize multi-label classification. The invention solves the problem that previous work on image data cannot simultaneously extract the secondary correlation and multiple correlation of tags, reduces the prediction difficulty caused by too sparse image data, and improves the accuracy of multi-label classification.

Description

technical field [0001] The invention relates to a neural network, a factorization machine and a multi-label classification method in the field of machine learning. Background technique [0002] With the rapid development of artificial intelligence, single-label classification is one of the most famous machine learning problems, where each instance is associated with a single label. However, in many real-world applications, an instance may be associated with multiple labels. For example, a piece of news might be about finance and politics, a video might be about government and policy, and an image might contain beaches and trees. Therefore, multi-label classification is an important research area. [0003] At present, people have done some research on multi-label classification and proposed many methods. The most critical challenge of multi-label classification is how to learn the correlation between labels. For this reason, it is proposed to extract one-time correlation,...

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): G06K9/62
CPCG06F18/241
Inventor 陈刚谌晨王皓波胡天磊陈珂寿黎但
Owner ZHEJIANG 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