Eureka AIR delivers breakthrough ideas for toughest innovation challenges, trusted by R&D personnel around the world.

Image sorting method and system for active learning

A technology of active learning and classification method, applied in character and pattern recognition, instruments, computer parts and other directions, can solve the problems of large sampling workload, time-consuming and laborious, achieve high uncertainty and representativeness, improve processing efficiency, The effect of reducing sample processing time and effort

Active Publication Date: 2014-03-05
SUZHOU UNIV
View PDF2 Cites 18 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, when the above method collects image samples, the uncertainty and representativeness of the samples are comprehensively considered, and the image sample with a larger combined value of the two is selected as a sample with higher information content. , for all samples in the original unlabeled sample set, it is necessary to calculate and measure the uncertainty and representativeness of each sample, resulting in a large sampling workload, time-consuming and laborious, especially when the unlabeled sample set is large, this shortcoming more prominent

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 sorting method and system for active learning
  • Image sorting method and system for active learning
  • Image sorting method and system for active learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0056] The embodiment of the present invention discloses an active learning image classification method, please refer to figure 1 , the method includes the following steps:

[0057] S1: Analyze and process the original unlabeled image sample set to obtain the most uncertain image sample set including at least one image sample, and each image sample in the most uncertain image sample set corresponds to a characterizing The first parameter of the degree of uncertainty of the Z image category, the parameter value of the first parameter satisfies the preset condition that characterizes the high uncertainty of the image sample, wherein the Z is a natural number greater than 1.

[0058] In this embodiment, the uncertainty and representativeness of the image samples are comprehensively considered, and the image samples with higher uncertainty and higher representativeness are taken as image samples with higher information content, that is, for the preset Z image categories, the most ...

Embodiment 2

[0110] Each of the image samples to be labeled (that is, each image sample in the representative image sample set) screened in the first embodiment above has both high uncertainty and high representativeness. In the second embodiment, based on the samples to be labeled After having the above two advantages, continue to optimize the samples to be labeled, and use the expected error rate reduction strategy to select the most informative samples from the samples to be labeled.

[0111] The core idea of ​​the expected error rate reduction strategy is: for each image sample in all candidate image samples, add the image sample to the labeled image sample set (that is, the labeled image sample set that trains the current image classifier) , and use the added labeled image sample set to update the current image classifier to obtain a new classifier; after that, use the new classifier to classify the remaining candidate image samples in the candidate image samples, and calculate the new...

Embodiment 3

[0127] Embodiment 3 of the present invention discloses an active learning image classification system, which corresponds to the active learning image classification methods in Embodiment 1 and Embodiment 2.

[0128] First, corresponding to the flow of the active learning image classification method in the first embodiment, the second embodiment discloses a structure of the active learning image classification system, please refer to Figure 6 , the system includes a first sampling module 100 , a second sampling module 200 , a labeling module 300 , a training module 400 and a classification module 500 .

[0129] The first sampling module 100 is configured to analyze and process the original unlabeled image sample set to obtain the most uncertain image sample set including at least one image sample, and each image sample in the most uncertain image sample set corresponds to one A first parameter that characterizes its degree of uncertainty relative to the preset Z image categori...

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 sorting method and system for active learning. The image sorting method comprises the following steps: for an original unmarked image sample set, firstly considering the uncertainty of all samples only, and acquiring all image samples with teh relatively high uncertainty from the original unmarked image sample set to form a most uncertain image sample set; then evaluating the representativeness of all the samples in the most uncertain image sample set, acquiring all the samples with the relatively high representativeness and forming a most representative image sample set; subsequently, carrying out marking and sorter training on the selected samples with the relatively high uncertainty and representativeness, and sorting target images by utilizing the trained sorter. Therefore, the image sorting method and system disclosed by the invention have the advantages that the mode of layering evaluation is adopted, firstly the samples are reduced and screened on the basis of uncertainty, then the representativeness evaluation is carried out on the uncertain image sample set with the relatively high uncertainty and sample-scale reduction, so that the uncertainty and the representativeness of the samples are guaranteed, the processing time and the workload for sampling are reduced and the processing efficiency is improved.

Description

technical field [0001] The invention belongs to the technical field of image classification in pattern recognition and machine learning, and in particular relates to an active learning image classification method and system. Background technique [0002] Image classification is an image processing method that distinguishes different types of objects according to different features reflected in image information, which is a very important research topic in the field of image processing. The key problem of image classification is to train a classifier model with high classification accuracy based on the labeled pictures. [0003] Due to the high cost of image labeling (manual labeling by domain experts is required), this field provides an image classification method based on active learning to reduce the workload of manual labeling by domain experts. This method selects as few samples as possible but with high information content for category labeling, and trains a classifier...

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): G06K9/66
Inventor 赵朋朋李承超吴健鲜学丰崔志明
Owner SUZHOU UNIV
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
Eureka Blog
Learn More
PatSnap group products