Method for acquiring target detection model by using scale histogram matching method

A technology of histogram matching and target detection, applied in the field of computer vision, can solve problems such as improving performance and reducing, and achieve the effect of improving performance, improving ability, and simple and effective matching technology

Inactive Publication Date: 2020-08-11
XI AN JIAOTONG UNIV
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, when the sample distribution of the task-specified data set is quite different from the sample distribution

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
  • Method for acquiring target detection model by using scale histogram matching method
  • Method for acquiring target detection model by using scale histogram matching method
  • Method for acquiring target detection model by using scale histogram matching method

Examples

Experimental program
Comparison scheme
Effect test

Example Embodiment

[0012] In one embodiment, such as figure 1 As shown, it discloses a method for obtaining a target detection model using a scale histogram matching method, and the method includes the following steps:

[0013] S100: Match the scale distribution in the public data set used for pre-training to the scale distribution in the target data set;

[0014] S200: Then use the public data set after scale matching to perform pre-training to obtain a preliminary target detection model;

[0015] S300: Use the preliminary target detection model to train on the target training set to obtain a final target detection model.

[0016] As far as this embodiment is concerned, with the emergence and development of convolutional neural networks, the research on visual inspection tasks has made unprecedented progress. However, how to train a detection model that meets actual requirements from a data set with a limited number of samples is still current Research the problems in the development process. In exper...

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 method for acquiring a target detection model by using a scale histogram matching method. The method comprises the following steps of: S100, matching scale distribution in apublic data set for pre-training into scale distribution in a target data set; S200, performing pre-training by using the public data set after scale matching to obtain a preliminary target detectionmodel; and S300, training on the target training set by using the preliminary target detection model to obtain a final target detection model. The method can effectively solve the problem that model performance improvement is limited when a public data set is used for pre-training due to the fact that target detection, especially related tasks in the industrial field, lacks enough sample data sets.

Description

technical field [0001] The disclosure belongs to computer vision, image processing and deep learning, and in particular relates to a scale histogram matching method for pre-training data sets. Background technique [0002] In recent years, deep learning has made significant progress in the field of computer vision, and convolutional neural networks have the characteristics of sparse interaction and parameter sharing, showing great advantages in the field of image detection. The excellent performance of convolutional neural networks also depends on the quality and size of the dataset associated with the task, regardless of the detection framework, the more data used for training, the better the performance of the detector. However, the cost of collecting data for specific tasks, especially in the industrial field, is very high, including a series of links such as data acquisition, data cleaning, and data labeling, which require a lot of manpower and material resources. Theref...

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
IPC IPC(8): G06T7/00G06N3/04G06N3/08
CPCG06T7/0002G06N3/08G06T2207/20081G06N3/045
Inventor 王小华杨爱军郭越祝金国王璐缙岳凡丁袁欢荣命哲
Owner XI AN JIAOTONG 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
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