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

End-to-end human body detection and attribute identification method

A human body detection and attribute recognition technology, applied in the field of target detection and human attribute recognition, can solve the problems of low model training efficiency and ineffective use of experience knowledge, achieve high recognition processing efficiency, reduce convolution calculations, and generalize performance. Good results

Active Publication Date: 2020-12-15
UNIV OF ELECTRONIC SCI & TECH OF CHINA
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to solve the problems of low model training efficiency and ineffective use of empirical knowledge in the existing human body attribute recognition methods. The present invention proposes an end-to-end human body detection and attribute recognition method based on multi-task 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
  • End-to-end human body detection and attribute identification method
  • End-to-end human body detection and attribute identification method
  • End-to-end human body detection and attribute identification method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0068] see figure 1 , the specific implementation process of this implementation includes:

[0069] Step 101: Obtain a detection data set including a human body. Collect multiple detection data sets containing human objects, delete samples that do not contain humans, and only keep samples containing human bodies, and unify the annotation format of multiple data sets, and randomly arrange them to obtain the data set DB 1 .

[0070] Step 102: Obtain the human body attribute recognition data set, and perform attribute alignment on multiple human body attribute recognition data sets, that is, take the union S of the attribute sets of all data sets, and use the attributes in the union S as the attributes of the integrated data set Attribute set, set default values ​​for missing attributes in the dataset, such as -1. And unify the annotation format of multiple datasets, and randomly arrange them to get the dataset DB 2 .

[0071] Step 103: The two data sets obtained in steps 10...

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 provides an end-to-end human body detection and attribute identification method based on deep learning, and aims to improve the generalization performance while improving the network operation efficiency. The network structure is composed of a target detection module and a human body attribute recognition module, and the target detection module completes recognition and positioning of a human body object. The human body attribute recognition module is a multi-output network and is used for completing judgment of a plurality of human body attributes. The model can accurately detect a plurality of persons in a real scene and detect attributes of the persons, and meanwhile, a method for guiding network training by taking attribute correlation as priori knowledge is set in combination with characteristics of the model.

Description

technical field [0001] The invention relates to the fields of target detection and human body attribute recognition, in particular to human body attribute recognition in real scenes. Background technique [0002] Human body attribute recognition refers to the judgment of human body attributes such as gender, age, hairstyle, and clothing in real scenes. These attributes have many applications in pedestrian recognition and retrieval. For example, when the video quality is poor, pedestrians are authenticated; in criminal investigation cases, similar suspects can be retrieved in surveillance videos through the external attributes of criminal suspects. [0003] The existing human attribute recognition method mainly regards target detection and attribute recognition as two independent tasks, builds a deep convolutional neural network for the two tasks to learn, and then connects the two networks in series. This type of method is relatively simple to implement, but Since the dete...

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): G06K9/00G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06V40/103G06V20/41G06N3/045G06F18/214
Inventor 陈爱国赵太银朱大勇罗光春谷俊霖杨栋栋
Owner UNIV OF ELECTRONIC SCI & TECH OF CHINA
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