Motion Primitives and Probabilistic Edit Distance for Action Recognition

Research output: Contribution to journalJournal articleResearchpeer-review

Abstract

The number of potential applications has made automatic recognition of human actions a very active research area. Different approaches have been followed based on trajectories through some state space. In this paper we also model an action as a trajectory through a state space, but we represent the actions as a sequence of temporal isolated instances, denoted primitives. These primitives are each defined by four features extracted from motion images. The primitives are recognized in each frame based on a trained classifier resulting in a sequence of primitives. From this sequence we recognize different temporal actions using a probabilistic Edit Distance method. The method is tested on different actions with and without noise and the results show recognition rates of 88.7% and 85.5%, respectively.
Original languageEnglish
Book seriesLecture Notes in Computer Science
Volume5085
Issue number1
Pages (from-to)24-35
Number of pages12
ISSN0302-9743
DOIs
Publication statusPublished - 2009

Fingerprint

Action Recognition
Edit Distance
Trajectories
Motion
Classifiers
State Space
Trajectory
Classifier

Keywords

  • Computer vision
  • Human action recognition
  • Motion primitives

Cite this

@article{0f5bc570eea511ddb0a4000ea68e967b,
title = "Motion Primitives and Probabilistic Edit Distance for Action Recognition",
abstract = "The number of potential applications has made automatic recognition of human actions a very active research area. Different approaches have been followed based on trajectories through some state space. In this paper we also model an action as a trajectory through a state space, but we represent the actions as a sequence of temporal isolated instances, denoted primitives. These primitives are each defined by four features extracted from motion images. The primitives are recognized in each frame based on a trained classifier resulting in a sequence of primitives. From this sequence we recognize different temporal actions using a probabilistic Edit Distance method. The method is tested on different actions with and without noise and the results show recognition rates of 88.7{\%} and 85.5{\%}, respectively.",
keywords = "Computer vision, Human action recognition, Motion primitives",
author = "Preben Fihl and Holte, {Michael Boelstoft} and Moeslund, {Thomas B.}",
year = "2009",
doi = "10.1007/978-3-540-92865-2_3",
language = "English",
volume = "5085",
pages = "24--35",
journal = "Lecture Notes in Computer Science",
issn = "0302-9743",
publisher = "Physica-Verlag",
number = "1",

}

Motion Primitives and Probabilistic Edit Distance for Action Recognition. / Fihl, Preben; Holte, Michael Boelstoft; Moeslund, Thomas B.

In: Lecture Notes in Computer Science, Vol. 5085, No. 1, 2009, p. 24-35.

Research output: Contribution to journalJournal articleResearchpeer-review

TY - JOUR

T1 - Motion Primitives and Probabilistic Edit Distance for Action Recognition

AU - Fihl, Preben

AU - Holte, Michael Boelstoft

AU - Moeslund, Thomas B.

PY - 2009

Y1 - 2009

N2 - The number of potential applications has made automatic recognition of human actions a very active research area. Different approaches have been followed based on trajectories through some state space. In this paper we also model an action as a trajectory through a state space, but we represent the actions as a sequence of temporal isolated instances, denoted primitives. These primitives are each defined by four features extracted from motion images. The primitives are recognized in each frame based on a trained classifier resulting in a sequence of primitives. From this sequence we recognize different temporal actions using a probabilistic Edit Distance method. The method is tested on different actions with and without noise and the results show recognition rates of 88.7% and 85.5%, respectively.

AB - The number of potential applications has made automatic recognition of human actions a very active research area. Different approaches have been followed based on trajectories through some state space. In this paper we also model an action as a trajectory through a state space, but we represent the actions as a sequence of temporal isolated instances, denoted primitives. These primitives are each defined by four features extracted from motion images. The primitives are recognized in each frame based on a trained classifier resulting in a sequence of primitives. From this sequence we recognize different temporal actions using a probabilistic Edit Distance method. The method is tested on different actions with and without noise and the results show recognition rates of 88.7% and 85.5%, respectively.

KW - Computer vision

KW - Human action recognition

KW - Motion primitives

U2 - 10.1007/978-3-540-92865-2_3

DO - 10.1007/978-3-540-92865-2_3

M3 - Journal article

VL - 5085

SP - 24

EP - 35

JO - Lecture Notes in Computer Science

JF - Lecture Notes in Computer Science

SN - 0302-9743

IS - 1

ER -