With the advent of smart devices and the massive use of smartphones, hand gesture recognition by mobile devices is being a major difficulty due to their technical specificities. To find a balance between speed and accuracy, we propose a new approach to recognize hand gestures by a smart device. This topic has some current interest and future applicability. In this paper, we present a new gesture detection framework for mobile devices based on LBP and SVM. LBP provides good texture representation properties. First, the proposed LBP on each non-overlapping blocks of a hand pose image is computed and a histogram of these LBPs is obtained. Those histograms are used as feature vectors for gesture classification as they demonstrate their robustness against compression and uniform intensity variations. The classification has been done by using Support Vector Machine (SVM). Since SVM is commonly used for pattern recognition, it is good for the explicit classification of form-dependent data, such as hand gestures. A recognition rate of approximately 93\% is obtained based the enhanced NUS database I. In addition, the impact of using such a hand pose estimation task in an embedded device is studied. We conduct experiments on the speed of detection on different mobile devices. The impact of using SVM as a classifier for a gesture recognition task in an embedded device like smartphone is studied.
Description
Hand Gesture Recognition System Based on LBP and SVM for Mobile Devices | SpringerLink
%0 Conference Paper
%1 lahiani2019gesture
%A Lahiani, Houssem
%A Neji, Mahmoud
%B Computational Collective Intelligence
%C Cham
%D 2019
%E Nguyen, Ngoc Thanh
%E Chbeir, Richard
%E Exposito, Ernesto
%E Aniorté, Philippe
%E Trawiński, Bogdan
%I Springer
%K android hand-gesture human-mobile-interaction lbp local-binary-patterns real support-vector-machine svm
%P 283-295
%R 10.1007/978-3-030-28377-3_23
%T Hand Gesture Recognition System Based on LBP and SVM for Mobile Devices
%U https://link.springer.com/chapter/10.1007/978-3-030-28377-3_23
%X With the advent of smart devices and the massive use of smartphones, hand gesture recognition by mobile devices is being a major difficulty due to their technical specificities. To find a balance between speed and accuracy, we propose a new approach to recognize hand gestures by a smart device. This topic has some current interest and future applicability. In this paper, we present a new gesture detection framework for mobile devices based on LBP and SVM. LBP provides good texture representation properties. First, the proposed LBP on each non-overlapping blocks of a hand pose image is computed and a histogram of these LBPs is obtained. Those histograms are used as feature vectors for gesture classification as they demonstrate their robustness against compression and uniform intensity variations. The classification has been done by using Support Vector Machine (SVM). Since SVM is commonly used for pattern recognition, it is good for the explicit classification of form-dependent data, such as hand gestures. A recognition rate of approximately 93\% is obtained based the enhanced NUS database I. In addition, the impact of using such a hand pose estimation task in an embedded device is studied. We conduct experiments on the speed of detection on different mobile devices. The impact of using SVM as a classifier for a gesture recognition task in an embedded device like smartphone is studied.
%@ 978-3-030-28377-3
@inproceedings{lahiani2019gesture,
abstract = {With the advent of smart devices and the massive use of smartphones, hand gesture recognition by mobile devices is being a major difficulty due to their technical specificities. To find a balance between speed and accuracy, we propose a new approach to recognize hand gestures by a smart device. This topic has some current interest and future applicability. In this paper, we present a new gesture detection framework for mobile devices based on LBP and SVM. LBP provides good texture representation properties. First, the proposed LBP on each non-overlapping blocks of a hand pose image is computed and a histogram of these LBPs is obtained. Those histograms are used as feature vectors for gesture classification as they demonstrate their robustness against compression and uniform intensity variations. The classification has been done by using Support Vector Machine (SVM). Since SVM is commonly used for pattern recognition, it is good for the explicit classification of form-dependent data, such as hand gestures. A recognition rate of approximately 93{\%} is obtained based the enhanced NUS database I. In addition, the impact of using such a hand pose estimation task in an embedded device is studied. We conduct experiments on the speed of detection on different mobile devices. The impact of using SVM as a classifier for a gesture recognition task in an embedded device like smartphone is studied.},
added-at = {2019-11-14T05:17:22.000+0100},
address = {Cham},
author = {Lahiani, Houssem and Neji, Mahmoud},
biburl = {https://www.bibsonomy.org/bibtex/286a535d0f333461c1d66c9ca7c399f9a/jpmor},
booktitle = {Computational Collective Intelligence},
day = 09,
description = {Hand Gesture Recognition System Based on LBP and SVM for Mobile Devices | SpringerLink},
doi = {10.1007/978-3-030-28377-3_23},
editor = {Nguyen, Ngoc Thanh and Chbeir, Richard and Exposito, Ernesto and Aniort{\'e}, Philippe and Trawi{\'{n}}ski, Bogdan},
interhash = {899426efa859fd8aab18ea808bd052b2},
intrahash = {86a535d0f333461c1d66c9ca7c399f9a},
isbn = {978-3-030-28377-3},
keywords = {android hand-gesture human-mobile-interaction lbp local-binary-patterns real support-vector-machine svm},
language = {English},
month = {08},
pages = {283-295},
publisher = {Springer},
timestamp = {2020-10-07T13:36:50.000+0200},
title = {Hand Gesture Recognition System Based on LBP and SVM for Mobile Devices},
url = {https://link.springer.com/chapter/10.1007/978-3-030-28377-3_23},
year = 2019
}