通过视频帧和光流运动模板使用 3D-CNN 和 2D-CNN 的双流融合模型进行手势识别。

IF 1.1 Q4 COMPUTER SCIENCE, SOFTWARE ENGINEERING
Debajit Sarma, V Kavyasree, M K Bhuyan
{"title":"通过视频帧和光流运动模板使用 3D-CNN 和 2D-CNN 的双流融合模型进行手势识别。","authors":"Debajit Sarma, V Kavyasree, M K Bhuyan","doi":"10.1007/s11334-022-00477-z","DOIUrl":null,"url":null,"abstract":"<p><p>Hand gestures are useful tools for many applications in the human-computer interaction community. Here, the objective is to track the movement of the hand irrespective of the shape, size and color of the hand. And, for this, a motion template guided by optical flow (OFMT) is proposed. OFMT is a compact representation of the motion information of a gesture encoded into a single image. Recently, deep networks have shown impressive improvements as compared to conventional hand-crafted feature-based techniques. Moreover, it is seen that the use of different streams with informative input data helps to increase the recognition performance. This work basically proposes a two-stream fusion model for hand gesture recognition. The two-stream network consists of two layers-a 3D convolutional neural network (C3D) that takes gesture videos as input and a 2D-CNN that takes OFMT images as input. C3D has shown its efficiency in capturing spatiotemporal information of a video, whereas OFMT helps to eliminate irrelevant gestures providing additional motion information. Though each stream can work independently, they are combined with a fusion scheme to boost the recognition results. We have shown the efficiency of the proposed two-stream network on two databases.</p>","PeriodicalId":44465,"journal":{"name":"Innovations in Systems and Software Engineering","volume":null,"pages":null},"PeriodicalIF":1.1000,"publicationDate":"2022-08-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9420673/pdf/","citationCount":"0","resultStr":"{\"title\":\"Two-stream fusion model using 3D-CNN and 2D-CNN via video-frames and optical flow motion templates for hand gesture recognition.\",\"authors\":\"Debajit Sarma, V Kavyasree, M K Bhuyan\",\"doi\":\"10.1007/s11334-022-00477-z\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p><p>Hand gestures are useful tools for many applications in the human-computer interaction community. Here, the objective is to track the movement of the hand irrespective of the shape, size and color of the hand. And, for this, a motion template guided by optical flow (OFMT) is proposed. OFMT is a compact representation of the motion information of a gesture encoded into a single image. Recently, deep networks have shown impressive improvements as compared to conventional hand-crafted feature-based techniques. Moreover, it is seen that the use of different streams with informative input data helps to increase the recognition performance. This work basically proposes a two-stream fusion model for hand gesture recognition. The two-stream network consists of two layers-a 3D convolutional neural network (C3D) that takes gesture videos as input and a 2D-CNN that takes OFMT images as input. C3D has shown its efficiency in capturing spatiotemporal information of a video, whereas OFMT helps to eliminate irrelevant gestures providing additional motion information. Though each stream can work independently, they are combined with a fusion scheme to boost the recognition results. We have shown the efficiency of the proposed two-stream network on two databases.</p>\",\"PeriodicalId\":44465,\"journal\":{\"name\":\"Innovations in Systems and Software Engineering\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":1.1000,\"publicationDate\":\"2022-08-29\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9420673/pdf/\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Innovations in Systems and Software Engineering\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1007/s11334-022-00477-z\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q4\",\"JCRName\":\"COMPUTER SCIENCE, SOFTWARE ENGINEERING\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Innovations in Systems and Software Engineering","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1007/s11334-022-00477-z","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"COMPUTER SCIENCE, SOFTWARE ENGINEERING","Score":null,"Total":0}
引用次数: 0

摘要

手势是人机交互领域许多应用的有用工具。在这里,我们的目标是跟踪手的运动,而不管手的形状、大小和颜色如何。为此,我们提出了一种由光流(OFMT)引导的运动模板。OFMT 是将手势的运动信息编码成单个图像的紧凑表示。最近,与传统的基于特征的手工技术相比,深度网络已经显示出令人印象深刻的改进。此外,使用不同的信息流输入数据有助于提高识别性能。这项工作主要提出了一种用于手势识别的双流融合模型。双流网络由两层组成,一层是以手势视频为输入的三维卷积神经网络(C3D),另一层是以 OFMT 图像为输入的二维卷积神经网络(2D-CNN)。C3D 在捕捉视频的时空信息方面表现出很高的效率,而 OFMT 则有助于消除无关的手势,提供额外的运动信息。虽然每个数据流都能独立工作,但它们通过融合方案结合在一起,以提高识别结果。我们在两个数据库上展示了所提出的双流网络的效率。
本文章由计算机程序翻译,如有差异,请以英文原文为准。

Two-stream fusion model using 3D-CNN and 2D-CNN via video-frames and optical flow motion templates for hand gesture recognition.

Two-stream fusion model using 3D-CNN and 2D-CNN via video-frames and optical flow motion templates for hand gesture recognition.

Two-stream fusion model using 3D-CNN and 2D-CNN via video-frames and optical flow motion templates for hand gesture recognition.

Two-stream fusion model using 3D-CNN and 2D-CNN via video-frames and optical flow motion templates for hand gesture recognition.

Hand gestures are useful tools for many applications in the human-computer interaction community. Here, the objective is to track the movement of the hand irrespective of the shape, size and color of the hand. And, for this, a motion template guided by optical flow (OFMT) is proposed. OFMT is a compact representation of the motion information of a gesture encoded into a single image. Recently, deep networks have shown impressive improvements as compared to conventional hand-crafted feature-based techniques. Moreover, it is seen that the use of different streams with informative input data helps to increase the recognition performance. This work basically proposes a two-stream fusion model for hand gesture recognition. The two-stream network consists of two layers-a 3D convolutional neural network (C3D) that takes gesture videos as input and a 2D-CNN that takes OFMT images as input. C3D has shown its efficiency in capturing spatiotemporal information of a video, whereas OFMT helps to eliminate irrelevant gestures providing additional motion information. Though each stream can work independently, they are combined with a fusion scheme to boost the recognition results. We have shown the efficiency of the proposed two-stream network on two databases.

求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
Innovations in Systems and Software Engineering
Innovations in Systems and Software Engineering COMPUTER SCIENCE, SOFTWARE ENGINEERING-
CiteScore
3.80
自引率
8.30%
发文量
75
期刊介绍: Innovations in Systems and Software Engineering: A NASA Journal addresses issues and innovations in Systems Engineering, Systems Integration, Software Engineering, Software Development and other related areas that are specifically of interest to NASA. The journal includes peer-reviewed world-class technical papers on topics of research, development and practice related to NASA''s missions and projects, topics of interest to NASA for future use, and topics describing problem areas for NASA together with potential solutions. Papers that do not address issues related to NASA are of course very welcome, provided that they address topics that NASA might like to consider for the future. Papers are solicited from NASA and government employees, contractors, NASA-supported academic and industrial partners, and non-NASA-supported academics and industrialists both in the USA and worldwide. The journal includes updates on NASA innovations, articles on NASA initiatives, papers looking at educational activities, and a State-of-the-Art section that gives an overview of specific topic areas in a comprehensive format written by an expert in the field.
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信