Learning to judge like a human: Convolutional networks for classification of ski jumping errors

Heike Brock, Yuji Ohgi, James Lee

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

Abstract

Advanced machine learning technologies are seldom applied to wearable motion sensor data obtained from sport movements. In this work, we therefore investigated neural networks for motion performance evaluation utilizing a set of inertial sensor-based ski jump measurements. A multidimensional convolutional network model that related the motion data under aspects of time, placement and sensor type was implemented. Additionally, its applicability as a measure for automatic motion style judging was evaluated. Results indicate that one multi-dimensional convolutional layer is sufficient to recognize relevant performance error representations. Furthermore, comparisons against a Support Vector Machine and a Hidden Markov Model show that the new model outperforms feature-based methods under noisy and biased data environments. Architectures such as the proposed evaluation system can hence become essential for automatic performance analysis and style judging systems in future.

Original languageEnglish
Title of host publicationISWC 2017 - Proceedings of the 2017 ACM International Symposium on Wearable Computers
PublisherAssociation for Computing Machinery
Pages106-113
Number of pages8
VolumePart F130534
ISBN (Electronic)9781450351881
DOIs
Publication statusPublished - 2017 Sep 11
Event29th ACM International Symposium on Wearable Computers, ISWC 2017 - Maui, United States
Duration: 2017 Sep 112017 Sep 15

Other

Other29th ACM International Symposium on Wearable Computers, ISWC 2017
CountryUnited States
CityMaui
Period17/9/1117/9/15

Fingerprint

Ski jumps
Sensors
Hidden Markov models
Sports
Support vector machines
Learning systems
Neural networks

Keywords

  • Convolutional neural networks
  • Motion analysis
  • Motion data
  • Specialized activity recognition

ASJC Scopus subject areas

  • Software
  • Hardware and Architecture
  • Computer Networks and Communications

Cite this

Brock, H., Ohgi, Y., & Lee, J. (2017). Learning to judge like a human: Convolutional networks for classification of ski jumping errors. In ISWC 2017 - Proceedings of the 2017 ACM International Symposium on Wearable Computers (Vol. Part F130534, pp. 106-113). Association for Computing Machinery. https://doi.org/10.1145/3123021.3123038

Learning to judge like a human : Convolutional networks for classification of ski jumping errors. / Brock, Heike; Ohgi, Yuji; Lee, James.

ISWC 2017 - Proceedings of the 2017 ACM International Symposium on Wearable Computers. Vol. Part F130534 Association for Computing Machinery, 2017. p. 106-113.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Brock, H, Ohgi, Y & Lee, J 2017, Learning to judge like a human: Convolutional networks for classification of ski jumping errors. in ISWC 2017 - Proceedings of the 2017 ACM International Symposium on Wearable Computers. vol. Part F130534, Association for Computing Machinery, pp. 106-113, 29th ACM International Symposium on Wearable Computers, ISWC 2017, Maui, United States, 17/9/11. https://doi.org/10.1145/3123021.3123038
Brock H, Ohgi Y, Lee J. Learning to judge like a human: Convolutional networks for classification of ski jumping errors. In ISWC 2017 - Proceedings of the 2017 ACM International Symposium on Wearable Computers. Vol. Part F130534. Association for Computing Machinery. 2017. p. 106-113 https://doi.org/10.1145/3123021.3123038
Brock, Heike ; Ohgi, Yuji ; Lee, James. / Learning to judge like a human : Convolutional networks for classification of ski jumping errors. ISWC 2017 - Proceedings of the 2017 ACM International Symposium on Wearable Computers. Vol. Part F130534 Association for Computing Machinery, 2017. pp. 106-113
@inproceedings{1116c83c3ed44802b84e6fbcdd6d2cf7,
title = "Learning to judge like a human: Convolutional networks for classification of ski jumping errors",
abstract = "Advanced machine learning technologies are seldom applied to wearable motion sensor data obtained from sport movements. In this work, we therefore investigated neural networks for motion performance evaluation utilizing a set of inertial sensor-based ski jump measurements. A multidimensional convolutional network model that related the motion data under aspects of time, placement and sensor type was implemented. Additionally, its applicability as a measure for automatic motion style judging was evaluated. Results indicate that one multi-dimensional convolutional layer is sufficient to recognize relevant performance error representations. Furthermore, comparisons against a Support Vector Machine and a Hidden Markov Model show that the new model outperforms feature-based methods under noisy and biased data environments. Architectures such as the proposed evaluation system can hence become essential for automatic performance analysis and style judging systems in future.",
keywords = "Convolutional neural networks, Motion analysis, Motion data, Specialized activity recognition",
author = "Heike Brock and Yuji Ohgi and James Lee",
year = "2017",
month = "9",
day = "11",
doi = "10.1145/3123021.3123038",
language = "English",
volume = "Part F130534",
pages = "106--113",
booktitle = "ISWC 2017 - Proceedings of the 2017 ACM International Symposium on Wearable Computers",
publisher = "Association for Computing Machinery",

}

TY - GEN

T1 - Learning to judge like a human

T2 - Convolutional networks for classification of ski jumping errors

AU - Brock, Heike

AU - Ohgi, Yuji

AU - Lee, James

PY - 2017/9/11

Y1 - 2017/9/11

N2 - Advanced machine learning technologies are seldom applied to wearable motion sensor data obtained from sport movements. In this work, we therefore investigated neural networks for motion performance evaluation utilizing a set of inertial sensor-based ski jump measurements. A multidimensional convolutional network model that related the motion data under aspects of time, placement and sensor type was implemented. Additionally, its applicability as a measure for automatic motion style judging was evaluated. Results indicate that one multi-dimensional convolutional layer is sufficient to recognize relevant performance error representations. Furthermore, comparisons against a Support Vector Machine and a Hidden Markov Model show that the new model outperforms feature-based methods under noisy and biased data environments. Architectures such as the proposed evaluation system can hence become essential for automatic performance analysis and style judging systems in future.

AB - Advanced machine learning technologies are seldom applied to wearable motion sensor data obtained from sport movements. In this work, we therefore investigated neural networks for motion performance evaluation utilizing a set of inertial sensor-based ski jump measurements. A multidimensional convolutional network model that related the motion data under aspects of time, placement and sensor type was implemented. Additionally, its applicability as a measure for automatic motion style judging was evaluated. Results indicate that one multi-dimensional convolutional layer is sufficient to recognize relevant performance error representations. Furthermore, comparisons against a Support Vector Machine and a Hidden Markov Model show that the new model outperforms feature-based methods under noisy and biased data environments. Architectures such as the proposed evaluation system can hence become essential for automatic performance analysis and style judging systems in future.

KW - Convolutional neural networks

KW - Motion analysis

KW - Motion data

KW - Specialized activity recognition

UR - http://www.scopus.com/inward/record.url?scp=85030464349&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85030464349&partnerID=8YFLogxK

U2 - 10.1145/3123021.3123038

DO - 10.1145/3123021.3123038

M3 - Conference contribution

AN - SCOPUS:85030464349

VL - Part F130534

SP - 106

EP - 113

BT - ISWC 2017 - Proceedings of the 2017 ACM International Symposium on Wearable Computers

PB - Association for Computing Machinery

ER -