Model and implementation of body movement recognition using Support Vector Machines and Finite State Machines with Cartesian coordinates input for gesture-based interaction

The growth in the use of gesture-based interaction in video games has highlighted the potential for the use of such interaction method for a wide range of applications. This paper presents the implementation of an enhanced model for gesture recognition as input method for software applications. The...

Full description

Saved in:
Bibliographic Details
Published inJournal of Computer Science & Technology Vol. 13; no. 2; pp. 69 - 75
Main Authors Bettio, Raphael W. de, Silva, Andre H.C, Heimfarth, Tales, Freire, Andre P, Sa, Alex G.C. de
Format Journal Article
LanguageEnglish
Spanish
Published La Plata Graduate Network of Argentine Universities with Computer Science Schools (RedUNCI) 01.10.2013
Universidad Nacional de la Plata, Journal of Computer Science and Technology
Postgraduate Office, School of Computer Science, Universidad Nacional de La Plata
Subjects
Accelerometers
Acceptance tests
Applications programs
Cartesian coordinates
Comparative analysis
Computer & video games
Electronic devices
Finite state machines
fsm
gesture
Gesture recognition
kinect
Motion perception
Product enhancement
Software
Support vector machines
svm
Technology application
Video games
Online AccessGet full text

Cover

More Information
Summary:The growth in the use of gesture-based interaction in video games has highlighted the potential for the use of such interaction method for a wide range of applications. This paper presents the implementation of an enhanced model for gesture recognition as input method for software applications. The model uses Support Vector Machines (SVM) and Finite State Machines (FSM) and the implementation was based on a Kinect[R] device. The model uses data input based on Cartesian coordinates. The use of Cartesian coordinates enables more flexibility to generalise the use of the model to different applications, when compared to related work encountered in the literature based on accelerometer devices for data input. The results showed that the use of SVM and FSM with Cartesian coordinates as input for gesture-based interaction is very promising. The success rate in gesture recognition was 98%, from a training corpus of 9 sets obtained by recording real users' gestures. A proof-of-concept implementation of the gesture recognition interaction was performed using the application Google Earth[R]. A preliminary acceptance evaluation with users indicated that the interaction with the system via the implementation reported was satisfactory. Keywords: Gesture, SVM, FSM, Kinect, Model.
ISSN:1666-6046
1666-6038