Skip to main content
main-content

Tipp

Weitere Artikel dieser Ausgabe durch Wischen aufrufen

01.10.2019 | APPLIED PROBLEMS | Ausgabe 4/2019

Pattern Recognition and Image Analysis 4/2019

A Lightweight Network Based on Pyramid Residual Module for Human Pose Estimation

Zeitschrift:
Pattern Recognition and Image Analysis > Ausgabe 4/2019
Autoren:
Bingkun Gao, Ke Ma, Hongbo Bi, Ling Wang
Wichtige Hinweise
https://static-content.springer.com/image/art%3A10.1134%2FS1054661819040023/MediaObjects/11493_2019_6033_Fig7_HTML.gif
Bingkun Gao, born in 1962, received his M.D and PhD in Harbin Engineering University, China, in 1995 and 2003, respectively. He worked as a PDF (PostDoc Fellow) in Control Science and Engineering at Northeastern University, China, in 2003–2006. He is also worked as a visiting scholar in Technische Univ. Clausthal, Germany, in 2007. Currently, he is a professor in School of Electrical Information Engineering in NorthEast Petroleum University. His main research interests comprise machine vision, intelligent control, information processing and information transmission, etc. He is the member of the Youth Work Committee of China Automation Society and the evaluation expert of National Ministry of Education degree and postgraduate education. Author of more than 70 papers.
https://static-content.springer.com/image/art%3A10.1134%2FS1054661819040023/MediaObjects/11493_2019_6033_Fig8_HTML.gif
Ke Ma, born in 1993, received her bachelor degree from the Taiyuan University of Technology, China, in 2016. Currently, she is a graduate student in NorthEast Petroleum University, China. Her research interests comprise the human pose estimation, object detection, and image recognition tasks, etc.
https://static-content.springer.com/image/art%3A10.1134%2FS1054661819040023/MediaObjects/11493_2019_6033_Fig9_HTML.gif
Hongbo Bi, born in 1979, received his bachelor degree and master degree in communications engineering from NorthEast Petroleum University, China, in 2001 and 2004, respectively. He received his PhD in 2013 from Beijing University of Posts and Telecommunications and worked as a PDF (PostDoc Fellow) in Harbin Engineering University in 2014–2017. He is also worked as a visiting scholar in University of Waterloo (Canada) in 2014–2015. Currently, he is an associate professor in School of Electrical Information Engineering in NorthEast Petroleum University. His main research interests focus on saliency detection, compressive sensing, deep learning, digital watermarking, signal processing, etc.
https://static-content.springer.com/image/art%3A10.1134%2FS1054661819040023/MediaObjects/11493_2019_6033_Fig10_HTML.gif
Ling Wang, born in 1992, received her bachelor degree from the Hebei University of Technology, China, in 2016. Currently, she is a graduate student in NorthEast Petroleum University, China. Her research interests comprise the human pose estimation, object detection, and image matching, etc.

Abstract

The human pose estimation is one of the most popular research fields. Its current accuracy is satisfactory in some cases, however, there exists a challenge for practical application due to the limited memory and computational efficiency in FPGAs and other hardware. We propose a lightweight module based on the pyramid residual module in this work. We change the convolution mode by using the depth-wise separable convolutions structure. Meanwhile, the channel split module and channel shuffle module are added to change the feature graph dimension. As a result, the parameters of the network are reduced effectively. We test the network on standard benchmarks MPII dataset, our method reduces about 50% of the training storage space while maintaining comparable accuracy. The complexity is simplified from 9 GFLOPs to 3 GFLOPs.

Bitte loggen Sie sich ein, um Zugang zu diesem Inhalt zu erhalten

Sie möchten Zugang zu diesem Inhalt erhalten? Dann informieren Sie sich jetzt über unsere Produkte:

Springer Professional "Wirtschaft+Technik"

Online-Abonnement

Mit Springer Professional "Wirtschaft+Technik" erhalten Sie Zugriff auf:

  • über 69.000 Bücher
  • über 500 Zeitschriften

aus folgenden Fachgebieten:

  • Automobil + Motoren
  • Bauwesen + Immobilien
  • Business IT + Informatik
  • Elektrotechnik + Elektronik
  • Energie + Umwelt
  • Finance + Banking
  • Management + Führung
  • Marketing + Vertrieb
  • Maschinenbau + Werkstoffe
  • Versicherung + Risiko

Testen Sie jetzt 30 Tage kostenlos.

Springer Professional "Wirtschaft"

Online-Abonnement

Mit Springer Professional "Wirtschaft" erhalten Sie Zugriff auf:

  • über 58.000 Bücher
  • über 300 Zeitschriften

aus folgenden Fachgebieten:

  • Bauwesen + Immobilien
  • Business IT + Informatik
  • Finance + Banking
  • Management + Führung
  • Marketing + Vertrieb
  • Versicherung + Risiko




Testen Sie jetzt 30 Tage kostenlos.

Springer Professional "Technik"

Online-Abonnement

Mit Springer Professional "Technik" erhalten Sie Zugriff auf:

  • über 50.000 Bücher
  • über 380 Zeitschriften

aus folgenden Fachgebieten:

  • Automobil + Motoren
  • Bauwesen + Immobilien
  • Business IT + Informatik
  • Elektrotechnik + Elektronik
  • Energie + Umwelt
  • Maschinenbau + Werkstoffe




Testen Sie jetzt 30 Tage kostenlos.

Literatur
Über diesen Artikel

Weitere Artikel der Ausgabe 4/2019

Pattern Recognition and Image Analysis 4/2019 Zur Ausgabe

MATHEMATICAL THEORY OF IMAGES AND SIGNALS REPRESENTING, PROCESSING, ANALYSIS, RECOGNITION AND UNDERSTANDING

Generalized Spectral-Analytical Method and Its Applications in Image Analysis and Pattern Recognition Problems

Premium Partner

    Bildnachweise