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

Human body attribute identification method based on deep learning

A technology of attribute recognition and deep learning, applied in the field of human attribute recognition based on deep learning, to achieve the effect of improving accuracy, reducing redundant problems, and avoiding waste of resources

Active Publication Date: 2016-11-16
北京小白世纪网络科技有限公司
View PDF5 Cites 40 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] One of the purposes of the present invention is to provide a human body attribute recognition method based on deep learning to solve the problem of how to locate human body parts more quickly and accurately under various interference factors such as viewing angle, posture, and occlusion in human body attribute recognition, and then The problem of attribute recognition combined with corresponding human body parts

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
  • Human body attribute identification method based on deep learning
  • Human body attribute identification method based on deep learning
  • Human body attribute identification method based on deep learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0061] A kind of human body attribute identification method based on deep learning that the present invention proposes, comprises the following steps:

[0062] (1) Randomly select 6W images from the publicly available labeled datasets for human detection and attribute recognition, of which 1W images are from Flickr and 5W images are from Taobao;

[0063] (2) Use the poselet detector trained by Bourdev et al. to detect the bounding box data of the main character and the poselet contained in the bounding box data on 6W training sets;

[0064] (3) Input the bounding box data into the RPN for processing to obtain candidate regions;

[0065] (4) Input the candidate area into Fast R-CNN for processing to obtain the RoI feature vector;

[0066] (5) Input the RoI feature vector into the fully connected layer, output the feature vector of poselet, the feature vector of poselet includes softmax vector and bbox regressor vector;

[0067] (6) Use the poselet feature vector to extract th...

Embodiment 2

[0072] A kind of human body attribute identification method based on deep learning that the present invention proposes, comprises the following steps:

[0073] (1) Randomly select 6W images from the publicly available labeled datasets for human detection and attribute recognition, of which 1W images are from Flickr and 5W images are from Taobao;

[0074] (2) Use the poselet detector trained by Bourdev et al. to detect the bounding box data of the main character and the poselet contained in the bounding box data on 6W training sets;

[0075] (3) Input the bounding box data into the RPN for processing to obtain candidate regions;

[0076] (4) Input the candidate area into Fast R-CNN for processing to obtain the RoI feature vector;

[0077] (5) Input the RoI feature vector into the fully connected layer, output the feature vector of poselet, the feature vector of poselet includes softmax vector and bbox regressor vector;

[0078] (6) Use the poselet feature vector to extract th...

Embodiment 3

[0083] A kind of human body attribute identification method based on deep learning that the present invention proposes, comprises the following steps:

[0084] (1) Randomly select 6W images from the publicly available labeled datasets for human body detection and attribute recognition, of which 1W images are from Flickr and 5W images are from Taobao;

[0085] (2) Use the poselet detector trained by Bourdev et al. to detect the bounding box data of the main character and the poselet contained in the bounding box data on 6W training sets;

[0086] (3) Input the bounding box data into the RPN for processing to obtain candidate regions;

[0087] (4) Input the candidate area into Fast R-CNN for processing to obtain the RoI feature vector;

[0088] (5) Input the RoI feature vector into the fully connected layer, output the feature vector of poselet, the feature vector of poselet includes softmax vector and bbox regressor vector;

[0089] (6) Use the poselet feature vector to extra...

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 human body attribute identification method based on deep learning. The human body attribute identification method comprises the steps of: step S1, constructing a human body part data set by using poselet detector; step S2, training poselet feature vectors according to the human body part data set; step S3, and training a human body attribute classifier by utilizing the poselet feature vectors through a convolutional neural network. The human body attribute identification method can position human body parts rapidly and accurately under the condition of various interference factors such as viewing angle, posture and occlusion in human body attribute identification, and further carries out attribute identification by combining with the corresponding human body part.

Description

technical field [0001] The present invention relates to the technical field of computer vision, in particular to a human body attribute recognition method based on deep learning. Background technique [0002] Attributes, as a visual feature that can be perceived by both computers and humans, have been widely used in many computer vision applications in recent years, especially in traditional object recognition and classification techniques using attributes as an intermediate layer. The use of attributes, in addition to being used as an intermediate layer for object description, recognition, and transfer learning, another major research direction is to focus on human-related attributes. The earliest researchers mainly obtained human-related attributes from facial features. In recent years, studies have proved that human attributes such as gender, age, and race can also obtain clues from other areas of the human body such as arms and legs. Not only that, the research scope of...

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): G06K9/00G06K9/46G06K9/62
CPCG06V40/10G06V10/56G06F18/241
Inventor 孙杲果邢爽杜强
Owner 北京小白世纪网络科技有限公司
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