Human Activity Recognition Process Using 3-D Posture Data

Risultato della ricerca: Article

133 Citazioni (Scopus)

Abstract

In this paper, we present a method for recognizing human activities using information sensed by an RGB-D camera, namely the Microsoft Kinect. Our approach is based on the estimation of some relevant joints of the human body by means of the Kinect; three different machine learning techniques, i.e., K-means clustering, support vector machines, and hidden Markov models, are combined to detect the postures involved while performing an activity, to classify them, and to model each activity as a spatiotemporal evolution of known postures. Experiments were performed on Kinect Activity Recognition Dataset, a new dataset, and on CAD-60, a public dataset. Experimental results show that our solution outperforms four relevant works based on RGB-D image fusion, hierarchical Maximum Entropy Markov Model, Markov Random Fields, and Eigenjoints, respectively. The performance we achieved, i.e., precision/recall of 77.3% and 76.7%, and the ability to recognize the activities in real time show promise for applied use.
Lingua originaleEnglish
pagine (da-a)586-597
Numero di pagine12
RivistaIEEE Transactions on Human-Machine Systems
Volume45
Stato di pubblicazionePublished - 2015

Fingerprint

Image fusion
Hidden Markov models
Support vector machines
Learning systems
Computer aided design
Entropy
Cameras
CAD
entropy
Experiments
experiment
ability
learning
performance

All Science Journal Classification (ASJC) codes

  • Human Factors and Ergonomics
  • Control and Systems Engineering
  • Signal Processing
  • Human-Computer Interaction
  • Computer Science Applications
  • Computer Networks and Communications
  • Artificial Intelligence

Cita questo

@article{3986ec9e98304304918732f781963d4f,
title = "Human Activity Recognition Process Using 3-D Posture Data",
abstract = "In this paper, we present a method for recognizing human activities using information sensed by an RGB-D camera, namely the Microsoft Kinect. Our approach is based on the estimation of some relevant joints of the human body by means of the Kinect; three different machine learning techniques, i.e., K-means clustering, support vector machines, and hidden Markov models, are combined to detect the postures involved while performing an activity, to classify them, and to model each activity as a spatiotemporal evolution of known postures. Experiments were performed on Kinect Activity Recognition Dataset, a new dataset, and on CAD-60, a public dataset. Experimental results show that our solution outperforms four relevant works based on RGB-D image fusion, hierarchical Maximum Entropy Markov Model, Markov Random Fields, and Eigenjoints, respectively. The performance we achieved, i.e., precision/recall of 77.3{\%} and 76.7{\%}, and the ability to recognize the activities in real time show promise for applied use.",
author = "Salvatore Gaglio and Marco Morana and {Lo Re}, Giuseppe and Salvatore Gaglio",
year = "2015",
language = "English",
volume = "45",
pages = "586--597",
journal = "IEEE Transactions on Human-Machine Systems",
issn = "2168-2291",
publisher = "IEEE Systems, Man, and Cybernetics Society",

}

TY - JOUR

T1 - Human Activity Recognition Process Using 3-D Posture Data

AU - Gaglio, Salvatore

AU - Morana, Marco

AU - Lo Re, Giuseppe

AU - Gaglio, Salvatore

PY - 2015

Y1 - 2015

N2 - In this paper, we present a method for recognizing human activities using information sensed by an RGB-D camera, namely the Microsoft Kinect. Our approach is based on the estimation of some relevant joints of the human body by means of the Kinect; three different machine learning techniques, i.e., K-means clustering, support vector machines, and hidden Markov models, are combined to detect the postures involved while performing an activity, to classify them, and to model each activity as a spatiotemporal evolution of known postures. Experiments were performed on Kinect Activity Recognition Dataset, a new dataset, and on CAD-60, a public dataset. Experimental results show that our solution outperforms four relevant works based on RGB-D image fusion, hierarchical Maximum Entropy Markov Model, Markov Random Fields, and Eigenjoints, respectively. The performance we achieved, i.e., precision/recall of 77.3% and 76.7%, and the ability to recognize the activities in real time show promise for applied use.

AB - In this paper, we present a method for recognizing human activities using information sensed by an RGB-D camera, namely the Microsoft Kinect. Our approach is based on the estimation of some relevant joints of the human body by means of the Kinect; three different machine learning techniques, i.e., K-means clustering, support vector machines, and hidden Markov models, are combined to detect the postures involved while performing an activity, to classify them, and to model each activity as a spatiotemporal evolution of known postures. Experiments were performed on Kinect Activity Recognition Dataset, a new dataset, and on CAD-60, a public dataset. Experimental results show that our solution outperforms four relevant works based on RGB-D image fusion, hierarchical Maximum Entropy Markov Model, Markov Random Fields, and Eigenjoints, respectively. The performance we achieved, i.e., precision/recall of 77.3% and 76.7%, and the ability to recognize the activities in real time show promise for applied use.

UR - http://hdl.handle.net/10447/103781

M3 - Article

VL - 45

SP - 586

EP - 597

JO - IEEE Transactions on Human-Machine Systems

JF - IEEE Transactions on Human-Machine Systems

SN - 2168-2291

ER -