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

Specific clothing image identification and detection method based on machine learning

A technology of image recognition and detection methods, applied in neural learning methods, character and pattern recognition, instruments, etc., can solve problems such as unrealizable, huge number, etc., to achieve the effect of speeding up

Inactive Publication Date: 2017-12-05
HARBIN INST OF TECH
View PDF4 Cites 44 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0007] The purpose of the present invention is to solve the problem that the existing search for specific types of clothing images cannot be realized through manual detection due to the huge number, and proposes a specific clothing image recognition and detection method based on machine learning

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
  • Specific clothing image identification and detection method based on machine learning
  • Specific clothing image identification and detection method based on machine learning
  • Specific clothing image identification and detection method based on machine learning

Examples

Experimental program
Comparison scheme
Effect test

specific Embodiment approach 1

[0045] Specific implementation mode 1: The specific process of a specific clothing image recognition and detection method based on machine learning in this implementation mode is as follows:

[0046]The present invention uses a convolutional neural network as the underlying image processing algorithm for image content recognition, and a more general multi-layer feed-forward neural network is used in the upper layer. In the image target detection, the basic idea based on Faster R-CNN is adopted, and the deep neural network is used to train and regress the coordinate points of the generated area to achieve area refinement, and then the target area and the entire picture are input into the target detection network for image recognition . In order to reduce the complexity of the network, the parameters of the original feature extraction network (convolutional neural network) in both the region generation network model and the target detection network model are shared.

[0047] St...

specific Embodiment approach 2

[0054] Specific embodiment two: the difference between this embodiment and specific embodiment one is that the image classification database is constructed in the step one by one, and the image in the image classification database is processed by a self-service sampling method to obtain the processed image; the specific process is:

[0055] Step 111, download the clothing image library from academic libraries such as Google Gallery, Baidu Gallery and ImageNet, and divide the images into three categories: specific clothing men's clothing (image samples of Arab men's clothing), specific clothing women's clothing (image samples of Arabic women's clothing) and general clothing The class constitutes a sample set, in which L (3140) samples of specific clothing women's clothing and ordinary clothing are taken, and S (140) samples are taken from specific clothing women's clothing and ordinary clothing samples for the test of specific clothing women's clothing and ordinary clothing sampl...

specific Embodiment approach 3

[0065] Embodiment 3: The difference between this embodiment and Embodiment 1 or 2 is that the convolutional neural network is constructed in the step 1 or 2, and the parameters in the convolutional neural network are optimized to obtain the optimized convolutional neural network; The specific process is:

[0066] Step 121, the multi-layer feedforward neural network has a large amount of parameters, and the original position information of the image is lost when converting the two-dimensional image information into one-dimensional data processing, so only the multi-layer feedforward neural network is used for image processing. The effect of classification is not particularly good, so the present invention adds a convolutional neural network before the multi-layer feedforward neural network. With the deepening of the number of convolutional layers, the training speed and the amount of memory required increase proportionally. When Set to four convolutional layers, and when each l...

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

A specific clothing image identification and detection method based on machine learning is provided. The invention relates to an image identification and detection method. The purpose of the invention is to solve the problems of the prior art that a manual detection way cannot realize searching of images of specific kinds of clothing because of a large quantity. The method includes the following processes: firstly, identifying image contents, and obtaining identified image contents; constructing an image classification database, processing images in the image classification database through a self-service sampling method, and obtaining processed images; obtaining an optimized convolution neural network; carrying out model integration; obtaining N individual learning devices, and combining the N learning devices through a simple ballot way; and secondly, utilizing a Faster R-CNN method to detect the identified image contents obtained in the first step. The specific clothing image identification and detection method is applied to the clothing image identification and detection field.

Description

technical field [0001] The invention relates to an image recognition and detection method. Background technique [0002] The current development of science and technology has redefined the form of e-commerce. Consumers use the online mall to compare prices while experiencing the experience of the offline mall, thus crossing the gap between the offline and online malls and realizing shopping anytime, anywhere. However, it is still a huge challenge to quickly search for one's favorite product from a large number of online product entries. Although some problems have been solved in some recent studies on text retrieval, there are still many challenges in areas such as clothing detection. These challenges include the following parts: [0003] (1) Most commodity items lack useful canonical tags that can be used for indexing [0004] (2) Compared with text retrieval, the concept of product retrieval when shopping is mostly visual [0005] (3) Mobile shopping requires fast and ...

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): G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06N3/045G06F18/2414
Inventor 李德志马铭李杰师鹏程徐誉靳登云
Owner HARBIN INST OF TECH
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