Sign Language Recognition via Skeleton-Aware Multi-Model Ensemble

Sign language is commonly used by deaf or mute people to communicate but requires extensive effort to master. It is usually performed with the fast yet delicate movement of hand gestures, body posture, and even facial expressions. Current Sign Language Recognition (SLR) methods usually extract featu...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Jiang, Songyao, Sun, Bin, Wang, Lichen, Bai, Yue, Li, Kunpeng, Fu, Yun
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page
container_issue
container_start_page
container_title
container_volume
creator Jiang, Songyao
Sun, Bin
Wang, Lichen
Bai, Yue
Li, Kunpeng
Fu, Yun
description Sign language is commonly used by deaf or mute people to communicate but requires extensive effort to master. It is usually performed with the fast yet delicate movement of hand gestures, body posture, and even facial expressions. Current Sign Language Recognition (SLR) methods usually extract features via deep neural networks and suffer overfitting due to limited and noisy data. Recently, skeleton-based action recognition has attracted increasing attention due to its subject-invariant and background-invariant nature, whereas skeleton-based SLR is still under exploration due to the lack of hand annotations. Some researchers have tried to use off-line hand pose trackers to obtain hand keypoints and aid in recognizing sign language via recurrent neural networks. Nevertheless, none of them outperforms RGB-based approaches yet. To this end, we propose a novel Skeleton Aware Multi-modal Framework with a Global Ensemble Model (GEM) for isolated SLR (SAM-SLR-v2) to learn and fuse multi-modal feature representations towards a higher recognition rate. Specifically, we propose a Sign Language Graph Convolution Network (SL-GCN) to model the embedded dynamics of skeleton keypoints and a Separable Spatial-Temporal Convolution Network (SSTCN) to exploit skeleton features. The skeleton-based predictions are fused with other RGB and depth based modalities by the proposed late-fusion GEM to provide global information and make a faithful SLR prediction. Experiments on three isolated SLR datasets demonstrate that our proposed SAM-SLR-v2 framework is exceedingly effective and achieves state-of-the-art performance with significant margins. Our code will be available at https://github.com/jackyjsy/SAM-SLR-v2
doi_str_mv 10.48550/arxiv.2110.06161
format Article
fullrecord <record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2110_06161</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2110_06161</sourcerecordid><originalsourceid>FETCH-LOGICAL-a671-6992f67b975a692f32605d84c5653124d949d324560c520a6ec63109e8695dec3</originalsourceid><addsrcrecordid>eNotz8tOwzAUBFBvWKDCB7DCP-Di5028jKrykFIh0e6jW_s2snAdlKYF_p5SWM1oFiMdxu6UnNvaOfmA41c6zbU6DxIUqGvWrFNfeIulP2JP_I3C0Jc0paHwU0K-fqdM01BE84kj8dUxT0mshkiZL8uB9ttMN-xqh_lAt_85Y5vH5WbxLNrXp5dF0wqESgnwXu-g2vrKIZyr0SBdrG1w4IzSNnrro9HWgQxOSwQKYJT0VIN3kYKZsfu_24uh-xjTHsfv7tfSXSzmB6ofQjc</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Sign Language Recognition via Skeleton-Aware Multi-Model Ensemble</title><source>arXiv.org</source><creator>Jiang, Songyao ; Sun, Bin ; Wang, Lichen ; Bai, Yue ; Li, Kunpeng ; Fu, Yun</creator><creatorcontrib>Jiang, Songyao ; Sun, Bin ; Wang, Lichen ; Bai, Yue ; Li, Kunpeng ; Fu, Yun</creatorcontrib><description>Sign language is commonly used by deaf or mute people to communicate but requires extensive effort to master. It is usually performed with the fast yet delicate movement of hand gestures, body posture, and even facial expressions. Current Sign Language Recognition (SLR) methods usually extract features via deep neural networks and suffer overfitting due to limited and noisy data. Recently, skeleton-based action recognition has attracted increasing attention due to its subject-invariant and background-invariant nature, whereas skeleton-based SLR is still under exploration due to the lack of hand annotations. Some researchers have tried to use off-line hand pose trackers to obtain hand keypoints and aid in recognizing sign language via recurrent neural networks. Nevertheless, none of them outperforms RGB-based approaches yet. To this end, we propose a novel Skeleton Aware Multi-modal Framework with a Global Ensemble Model (GEM) for isolated SLR (SAM-SLR-v2) to learn and fuse multi-modal feature representations towards a higher recognition rate. Specifically, we propose a Sign Language Graph Convolution Network (SL-GCN) to model the embedded dynamics of skeleton keypoints and a Separable Spatial-Temporal Convolution Network (SSTCN) to exploit skeleton features. The skeleton-based predictions are fused with other RGB and depth based modalities by the proposed late-fusion GEM to provide global information and make a faithful SLR prediction. Experiments on three isolated SLR datasets demonstrate that our proposed SAM-SLR-v2 framework is exceedingly effective and achieves state-of-the-art performance with significant margins. Our code will be available at https://github.com/jackyjsy/SAM-SLR-v2</description><identifier>DOI: 10.48550/arxiv.2110.06161</identifier><language>eng</language><subject>Computer Science - Artificial Intelligence ; Computer Science - Computer Vision and Pattern Recognition</subject><creationdate>2021-10</creationdate><rights>http://creativecommons.org/licenses/by/4.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,776,881</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2110.06161$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2110.06161$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Jiang, Songyao</creatorcontrib><creatorcontrib>Sun, Bin</creatorcontrib><creatorcontrib>Wang, Lichen</creatorcontrib><creatorcontrib>Bai, Yue</creatorcontrib><creatorcontrib>Li, Kunpeng</creatorcontrib><creatorcontrib>Fu, Yun</creatorcontrib><title>Sign Language Recognition via Skeleton-Aware Multi-Model Ensemble</title><description>Sign language is commonly used by deaf or mute people to communicate but requires extensive effort to master. It is usually performed with the fast yet delicate movement of hand gestures, body posture, and even facial expressions. Current Sign Language Recognition (SLR) methods usually extract features via deep neural networks and suffer overfitting due to limited and noisy data. Recently, skeleton-based action recognition has attracted increasing attention due to its subject-invariant and background-invariant nature, whereas skeleton-based SLR is still under exploration due to the lack of hand annotations. Some researchers have tried to use off-line hand pose trackers to obtain hand keypoints and aid in recognizing sign language via recurrent neural networks. Nevertheless, none of them outperforms RGB-based approaches yet. To this end, we propose a novel Skeleton Aware Multi-modal Framework with a Global Ensemble Model (GEM) for isolated SLR (SAM-SLR-v2) to learn and fuse multi-modal feature representations towards a higher recognition rate. Specifically, we propose a Sign Language Graph Convolution Network (SL-GCN) to model the embedded dynamics of skeleton keypoints and a Separable Spatial-Temporal Convolution Network (SSTCN) to exploit skeleton features. The skeleton-based predictions are fused with other RGB and depth based modalities by the proposed late-fusion GEM to provide global information and make a faithful SLR prediction. Experiments on three isolated SLR datasets demonstrate that our proposed SAM-SLR-v2 framework is exceedingly effective and achieves state-of-the-art performance with significant margins. Our code will be available at https://github.com/jackyjsy/SAM-SLR-v2</description><subject>Computer Science - Artificial Intelligence</subject><subject>Computer Science - Computer Vision and Pattern Recognition</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2021</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotz8tOwzAUBFBvWKDCB7DCP-Di5028jKrykFIh0e6jW_s2snAdlKYF_p5SWM1oFiMdxu6UnNvaOfmA41c6zbU6DxIUqGvWrFNfeIulP2JP_I3C0Jc0paHwU0K-fqdM01BE84kj8dUxT0mshkiZL8uB9ttMN-xqh_lAt_85Y5vH5WbxLNrXp5dF0wqESgnwXu-g2vrKIZyr0SBdrG1w4IzSNnrro9HWgQxOSwQKYJT0VIN3kYKZsfu_24uh-xjTHsfv7tfSXSzmB6ofQjc</recordid><startdate>20211012</startdate><enddate>20211012</enddate><creator>Jiang, Songyao</creator><creator>Sun, Bin</creator><creator>Wang, Lichen</creator><creator>Bai, Yue</creator><creator>Li, Kunpeng</creator><creator>Fu, Yun</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20211012</creationdate><title>Sign Language Recognition via Skeleton-Aware Multi-Model Ensemble</title><author>Jiang, Songyao ; Sun, Bin ; Wang, Lichen ; Bai, Yue ; Li, Kunpeng ; Fu, Yun</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a671-6992f67b975a692f32605d84c5653124d949d324560c520a6ec63109e8695dec3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2021</creationdate><topic>Computer Science - Artificial Intelligence</topic><topic>Computer Science - Computer Vision and Pattern Recognition</topic><toplevel>online_resources</toplevel><creatorcontrib>Jiang, Songyao</creatorcontrib><creatorcontrib>Sun, Bin</creatorcontrib><creatorcontrib>Wang, Lichen</creatorcontrib><creatorcontrib>Bai, Yue</creatorcontrib><creatorcontrib>Li, Kunpeng</creatorcontrib><creatorcontrib>Fu, Yun</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Jiang, Songyao</au><au>Sun, Bin</au><au>Wang, Lichen</au><au>Bai, Yue</au><au>Li, Kunpeng</au><au>Fu, Yun</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Sign Language Recognition via Skeleton-Aware Multi-Model Ensemble</atitle><date>2021-10-12</date><risdate>2021</risdate><abstract>Sign language is commonly used by deaf or mute people to communicate but requires extensive effort to master. It is usually performed with the fast yet delicate movement of hand gestures, body posture, and even facial expressions. Current Sign Language Recognition (SLR) methods usually extract features via deep neural networks and suffer overfitting due to limited and noisy data. Recently, skeleton-based action recognition has attracted increasing attention due to its subject-invariant and background-invariant nature, whereas skeleton-based SLR is still under exploration due to the lack of hand annotations. Some researchers have tried to use off-line hand pose trackers to obtain hand keypoints and aid in recognizing sign language via recurrent neural networks. Nevertheless, none of them outperforms RGB-based approaches yet. To this end, we propose a novel Skeleton Aware Multi-modal Framework with a Global Ensemble Model (GEM) for isolated SLR (SAM-SLR-v2) to learn and fuse multi-modal feature representations towards a higher recognition rate. Specifically, we propose a Sign Language Graph Convolution Network (SL-GCN) to model the embedded dynamics of skeleton keypoints and a Separable Spatial-Temporal Convolution Network (SSTCN) to exploit skeleton features. The skeleton-based predictions are fused with other RGB and depth based modalities by the proposed late-fusion GEM to provide global information and make a faithful SLR prediction. Experiments on three isolated SLR datasets demonstrate that our proposed SAM-SLR-v2 framework is exceedingly effective and achieves state-of-the-art performance with significant margins. Our code will be available at https://github.com/jackyjsy/SAM-SLR-v2</abstract><doi>10.48550/arxiv.2110.06161</doi><oa>free_for_read</oa></addata></record>
fulltext fulltext_linktorsrc
identifier DOI: 10.48550/arxiv.2110.06161
ispartof
issn
language eng
recordid cdi_arxiv_primary_2110_06161
source arXiv.org
subjects Computer Science - Artificial Intelligence
Computer Science - Computer Vision and Pattern Recognition
title Sign Language Recognition via Skeleton-Aware Multi-Model Ensemble
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-03T09%3A18%3A17IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Sign%20Language%20Recognition%20via%20Skeleton-Aware%20Multi-Model%20Ensemble&rft.au=Jiang,%20Songyao&rft.date=2021-10-12&rft_id=info:doi/10.48550/arxiv.2110.06161&rft_dat=%3Carxiv_GOX%3E2110_06161%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true