Image classification method and system based on migration learning

A technology of transfer learning and classification methods, which is applied in the fields of image classification, target recognition, target retrieval, and database management. It can solve problems such as fitting and local optimal solutions, so as to improve classification accuracy, avoid over-fitting and local optimal solutions problem-solving effect

Inactive Publication Date: 2018-10-12
HUBEI UNIV OF TECH
View PDF3 Cites 29 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, in the deep learning algorithm, if the amount of data is too small and imported into the complex model, there will be problems of overfitting and local optimal solution

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
  • Image classification method and system based on migration learning
  • Image classification method and system based on migration learning
  • Image classification method and system based on migration learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0055] The technical solution of the present invention will be further described below in conjunction with the accompanying drawings.

[0056] Step 1: Use the known training set A with similar features to make the training set B of the migration network through the support vector machine.

[0057] Step 1a: If figure 1 As shown in the process, to extract the HOG features of the training set A, firstly convert the RGB image to grayscale, and use the Gamma correction method to normalize the input image, and then use a set of formulas (1) to calculate the value of each pixel of the image Gradient (magnitude and direction). Then divide the image into 6*6 cells, and count the gradient histogram of each cell, use the detection window to divide the block block, detect 3*3 cells to form a block, if the division is not enough, it will appear Overlap, in the above formula, x and y are the pixel coordinates, and the gradient histogram of each cell is counted, and every 3*3 cells form a ...

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 an image classification method and system based on migration learning. The method comprises the steps that 1, a feature similarity known training set A is utilized to make a training set B of a migration network through a support vector machine; 2, a migration learning network is constructed; 3, the training set B classified in the step 1 is used as a training learning setof the migration learning network, and a migration learning network model with high robustness and good accuracy is obtained through training; and 4, a to-be-classified dataset is introduced into thetrained migration learning network model, and a final classification result is obtained and marked with a tag. Through the image classification method and system, the requirement that a big sample dataset is needed to serve as input when a common RGB image is trained through deep learning is overcome, the problems of overfitting and locally optimal solutions in the training process are avoided, and classification precision is improved to some extent compared with a traditional classification algorithm.

Description

technical field [0001] The invention belongs to the technical field of image classification, and is suitable for classification scenarios where few training samples are required for data to be classified or only samples with similar characteristics are required, and can be used in the fields of target recognition, target retrieval, database management and the like. Background technique [0002] In recent years, deep learning algorithms have been widely used in traditional RGB image classification, medical images and synthetic aperture radar (SAR) images where features are difficult to extract. With the improvement of the level of science and technology, new image data information is being further understood by humans, resulting in the possibility of failure of traditional image data sets and the limited number of new image data sets. Research on how to build a bridge between traditional image data sets and newly acquired image data is of great significance for realizing imag...

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): G06K9/62
CPCG06F18/2413G06F18/2411G06F18/24147
Inventor 王云艳罗冷坤徐超
Owner HUBEI UNIV OF TECH
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