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

A bottom-up multi-person attitude estimation method using bounding box constraints

A pose estimation, bottom-up technique, applied in the field of neural networks, it can solve the problems of increasing time complexity, unable to obtain ideal results, etc., to avoid error propagation, solve pose truncated, and achieve good accuracy and running time.

Inactive Publication Date: 2019-02-15
ZHEJIANG UNIV
View PDF4 Cites 13 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

But both methods have their own shortcomings: the top-down method has high requirements on the accuracy of the human detector, and the time complexity increases linearly with the number of people in the picture; the bottom-up method is severely occluded Unable to get the desired result in the case of

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
  • A bottom-up multi-person attitude estimation method using bounding box constraints
  • A bottom-up multi-person attitude estimation method using bounding box constraints
  • A bottom-up multi-person attitude estimation method using bounding box constraints

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0055] A bottom-up approach to multi-person pose estimation using bounding box constraints, including the following steps:

[0056] (1) Bounding box detection: Use YOLOv2 (J.Redmon and A.Farhadi, "Yolo9000: better, faster, stronger," arXiv preprint arXiv:1612.08242, 2016.) as a human detector to obtain the bounding box of the person in the picture B i ;

[0057] (2) Obtain network output: send the picture into the neural network we designed to obtain the confidence map and direction field information of the picture, and the neural network is obtained by the following methods:

[0058] Obtain training samples from the data set, take pictures as input, and use the confidence map S of 14 joints corresponding to each picture j and 13 directional fields L c As output, j=1,2,…,14; c=1,2,…,13, for neural network training network structure in Z.Cao, T.Simon, S.-E.Wei, and Y.Sheikh, "Realtime multi-person2d poseestimation using part affinity fields," arXiv preprint arXiv:1611.08050...

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 a bottom-up multi-person attitude estimation method using bounding box constraints. The main process is as follows: firstly, the existing multi-person attitude estimation neuralnetwork is improved again, so that the accuracy is higher; secondly, a greedy algorithm using bounding box is designed for pose resolution; in addition, a non-maximal suppression attitude detection and removal algorithm is designed to remove the duplicate results, and finally, a simple greedy missing nodes algorithm is used to make the result more accurate.

Description

technical field [0001] The present invention proposes an improved neural network, which can obtain more accurate results. Secondly, aiming at the problems existing in the existing methods of multi-person pose estimation due to factors such as occlusion and complex poses, a post-processing method using bounding boxes is proposed. Process the algorithm to ensure the correctness of the results. Background technique [0002] Multi-person human pose estimation is a very challenging task in the field of computer vision. Multi-person pose estimation aims to find out the skeletal keypoints of all persons in an image. Pose estimation for multiple people outdoors is very challenging due to the high flexibility of body poses, ego and external occlusions, different clothes, rare poses, etc. [0003] Due to the emergence of deep convolutional neural networks, existing multi-person pose estimation can be roughly divided into two categories: bottom-up methods and top-down methods. The t...

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): G06T7/00G06T7/13G06T7/70G06K9/00G06N3/04
CPCG06T7/0002G06T7/13G06T7/70G06T2207/10004G06T2207/20081G06V40/10G06N3/045
Inventor 刘新国李妙鹏周子孟
Owner ZHEJIANG UNIV
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