Automatic classification of sports movements through inertial sensors: a novel application to functional fitness

Stefano Nodari, Nicola Lopomo, Ezio Preatoni

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

Abstract

We aimed to develop and validate a method for the automatic detection and classification of four popular functional fitness drills. Five inertia measurement units were located on the upper and lower limb and on the trunk of fourteen participants. Accelerations and angular velocities were acquired continuously from the units, and used to train machine learning algorithms. Video footage was used for validation. Support vector machine with a cubic kernel gave the best classification performances (97.9% accuracy). The method never mistook a fitness movement for another, but confused relatively frequently (4-17%) a fitness movement phase with the transition between subsequent repetitions of the same task or between different drills. The proposed approach showed good potential for future on-field application, using currently available technologies such as smart- watches/phones.
Original languageEnglish
Title of host publicationProceedings of the 37th International Conference on Biomechanics in Sports (2019) Oxford (Ohio), USA, July 21-25, 2019
Number of pages4
Volume1
Publication statusAccepted/In press - 13 Mar 2019
Event37th International Conference on Biomechanics in Sports - Miami University, Ohio, Oxford, Ohio, USA United States
Duration: 21 Jul 201925 Jul 2019

Conference

Conference37th International Conference on Biomechanics in Sports
Abbreviated titleISBS2019
CountryUSA United States
CityOxford, Ohio
Period21/07/1925/07/19

Cite this

Nodari, S., Lopomo, N., & Preatoni, E. (Accepted/In press). Automatic classification of sports movements through inertial sensors: a novel application to functional fitness. In Proceedings of the 37th International Conference on Biomechanics in Sports (2019) Oxford (Ohio), USA, July 21-25, 2019 (Vol. 1)

Automatic classification of sports movements through inertial sensors: a novel application to functional fitness. / Nodari, Stefano; Lopomo, Nicola; Preatoni, Ezio.

Proceedings of the 37th International Conference on Biomechanics in Sports (2019) Oxford (Ohio), USA, July 21-25, 2019. Vol. 1 2019.

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

Nodari, S, Lopomo, N & Preatoni, E 2019, Automatic classification of sports movements through inertial sensors: a novel application to functional fitness. in Proceedings of the 37th International Conference on Biomechanics in Sports (2019) Oxford (Ohio), USA, July 21-25, 2019. vol. 1, 37th International Conference on Biomechanics in Sports, Oxford, Ohio, USA United States, 21/07/19.
Nodari S, Lopomo N, Preatoni E. Automatic classification of sports movements through inertial sensors: a novel application to functional fitness. In Proceedings of the 37th International Conference on Biomechanics in Sports (2019) Oxford (Ohio), USA, July 21-25, 2019. Vol. 1. 2019
Nodari, Stefano ; Lopomo, Nicola ; Preatoni, Ezio. / Automatic classification of sports movements through inertial sensors: a novel application to functional fitness. Proceedings of the 37th International Conference on Biomechanics in Sports (2019) Oxford (Ohio), USA, July 21-25, 2019. Vol. 1 2019.
@inproceedings{16d318319fd94ea4b9bf9fa23433afd7,
title = "Automatic classification of sports movements through inertial sensors: a novel application to functional fitness",
abstract = "We aimed to develop and validate a method for the automatic detection and classification of four popular functional fitness drills. Five inertia measurement units were located on the upper and lower limb and on the trunk of fourteen participants. Accelerations and angular velocities were acquired continuously from the units, and used to train machine learning algorithms. Video footage was used for validation. Support vector machine with a cubic kernel gave the best classification performances (97.9{\%} accuracy). The method never mistook a fitness movement for another, but confused relatively frequently (4-17{\%}) a fitness movement phase with the transition between subsequent repetitions of the same task or between different drills. The proposed approach showed good potential for future on-field application, using currently available technologies such as smart- watches/phones.",
author = "Stefano Nodari and Nicola Lopomo and Ezio Preatoni",
year = "2019",
month = "3",
day = "13",
language = "English",
volume = "1",
booktitle = "Proceedings of the 37th International Conference on Biomechanics in Sports (2019) Oxford (Ohio), USA, July 21-25, 2019",

}

TY - GEN

T1 - Automatic classification of sports movements through inertial sensors: a novel application to functional fitness

AU - Nodari, Stefano

AU - Lopomo, Nicola

AU - Preatoni, Ezio

PY - 2019/3/13

Y1 - 2019/3/13

N2 - We aimed to develop and validate a method for the automatic detection and classification of four popular functional fitness drills. Five inertia measurement units were located on the upper and lower limb and on the trunk of fourteen participants. Accelerations and angular velocities were acquired continuously from the units, and used to train machine learning algorithms. Video footage was used for validation. Support vector machine with a cubic kernel gave the best classification performances (97.9% accuracy). The method never mistook a fitness movement for another, but confused relatively frequently (4-17%) a fitness movement phase with the transition between subsequent repetitions of the same task or between different drills. The proposed approach showed good potential for future on-field application, using currently available technologies such as smart- watches/phones.

AB - We aimed to develop and validate a method for the automatic detection and classification of four popular functional fitness drills. Five inertia measurement units were located on the upper and lower limb and on the trunk of fourteen participants. Accelerations and angular velocities were acquired continuously from the units, and used to train machine learning algorithms. Video footage was used for validation. Support vector machine with a cubic kernel gave the best classification performances (97.9% accuracy). The method never mistook a fitness movement for another, but confused relatively frequently (4-17%) a fitness movement phase with the transition between subsequent repetitions of the same task or between different drills. The proposed approach showed good potential for future on-field application, using currently available technologies such as smart- watches/phones.

UR - https://isbs.org/isbs-conference

M3 - Conference contribution

VL - 1

BT - Proceedings of the 37th International Conference on Biomechanics in Sports (2019) Oxford (Ohio), USA, July 21-25, 2019

ER -