Image feature extraction method for garment image retrieval

A technology of image features and image retrieval, applied in still image data retrieval, neural learning methods, still image data query, etc., can solve the problem of low accuracy of Top1

Inactive Publication Date: 2019-01-01
李峰 +1
View PDF0 Cites 10 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0010] The purpose of the present invention is to overcome the problem of low Top1 accuracy rate of clothing retrieval caused by insufficient description of local information in the above-mentioned prior a

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 feature extraction method for garment image retrieval
  • Image feature extraction method for garment image retrieval
  • Image feature extraction method for garment image retrieval

Examples

Experimental program
Comparison scheme
Effect test

Example Embodiment

[0037] An image feature extraction method for clothing image retrieval is characterized by the following steps:

[0038] (1) Design as figure 1 The deep learning model shown in the figure; the light blue filled box in the figure represents the data, the number in it represents the dimension of the data, the green frame represents the operation on the data, the arrow represents the data flow, the red part represents the loss function; the input of the network contains images Data, image key point position collection and image attribute information collection, respectively denoted as B, P, A; B represents the original data of the image, which can be regarded as a three-dimensional matrix, and the three-dimensional represents the number of image channels, image height and image width respectively; P= {P 1 ,..., P i ,..., P m } Represents the set of image key point coordinates, where m represents the number of key points, P i =(x i , Y i ) Represents the position of the i-th key point...

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 relates to an image feature extraction method for garment image retrieval, belonging to the technical field of image retrieval. Firstly, a key region generation network based on key points and a key region fusion network for fusing global features and key region features of garment images are designed creatively. Then the key region generation network and the key region fusion network are added to the open source depth learning model VGG16 to obtain the depth learning model for clothing image retrieval. Then the model is converged by cross-training the key regions to generate thenetwork and the key regions to converge the network. Finally, the high-level features of the depth learning model proposed by the invention are extracted for the garment image retrieval task. The depth feature extraction method of the garment image provided by the invention can effectively improve the accuracy of garment image retrieval, and the method is simple and easy to realize.

Description

technical field [0001] The invention relates to the technical field of image retrieval, in particular to an image feature extraction method for clothing image retrieval. Background technique [0002] In recent years, with the continuous popularization and development of the Internet, especially the mobile Internet, people's lives have undergone rapid changes. In the past, the Internet information that people obtained was mainly text information, but now there is a huge demand for multimedia information such as images and videos. How to quickly and accurately find the information people need from a large amount of image data is becoming more and more important . [0003] There are currently many researches in the field of image retrieval. In [1], Wan et al. verified the effectiveness of the high-level features of the deep classification model for retrieval problems and their superiority over traditional features through experiments. In [2], Tolias et al. extracted the depth...

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): G06F16/53G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06N3/045G06F18/2414
Inventor 李峰白宇王斌旭
Owner 李峰
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