AI-driven form analysis: Personalised strength-training feedback via computer vision techniques

Sasha Westlake; Jarutas Andritsch; Drishty Sobnath

AI-driven form analysis: Personalised strength-training feedback via computer vision techniques

Sasha Westlake, Jarutas Andritsch, Drishty Sobnath

School of Mathematical & Computer Sciences

Research output: Contribution to conference › Paper › peer-review

Abstract

This study explores AI-driven form analysis in strength training, leveraging well-known techniques such as Convolutional Neural Networks (CNN) and Random Forest. The CNN model achieved an impressive 90.9% accuracy on test data, demonstrating resilience to noise, while the Random Forest model surpassed success criteria with 87.8% accuracy. The research highlights the potential of AI to enhance personal fitness training, making it safer and reducing injury risks, supporting the continued growth of AI-based solutions in the fitness industry.

Original language	English
Publication status	Published - 8 Jan 2024
Event	International Conference on Artificial Intelligence for Healthcare 2024 - London, United Kingdom Duration: 8 Jan 2024 → 9 Jan 2024

Conference

Conference	International Conference on Artificial Intelligence for Healthcare 2024
Abbreviated title	ICAIH-24
Country/Territory	United Kingdom
City	London
Period	8/01/24 → 9/01/24

Keywords

Artificial Intelligence
Computer Vision
Fitness
Health

Cite this

@conference{e6adc100c6dd4e9d9f9c99ea4a50f621,

title = "AI-driven form analysis: Personalised strength-training feedback via computer vision techniques",

abstract = "This study explores AI-driven form analysis in strength training, leveraging well-known techniques such as Convolutional Neural Networks (CNN) and Random Forest. The CNN model achieved an impressive 90.9% accuracy on test data, demonstrating resilience to noise, while the Random Forest model surpassed success criteria with 87.8% accuracy. The research highlights the potential of AI to enhance personal fitness training, making it safer and reducing injury risks, supporting the continued growth of AI-based solutions in the fitness industry.",

keywords = "Artificial Intelligence, Computer Vision, Fitness, Health",

author = "Sasha Westlake and Jarutas Andritsch and Drishty Sobnath",

year = "2024",

month = jan,

day = "8",

language = "English",

note = "International Conference on Artificial Intelligence for Healthcare 2024, ICAIH-24 ; Conference date: 08-01-2024 Through 09-01-2024",

}

TY - CONF

T1 - AI-driven form analysis: Personalised strength-training feedback via computer vision techniques

AU - Westlake, Sasha

AU - Andritsch, Jarutas

AU - Sobnath, Drishty

PY - 2024/1/8

Y1 - 2024/1/8

N2 - This study explores AI-driven form analysis in strength training, leveraging well-known techniques such as Convolutional Neural Networks (CNN) and Random Forest. The CNN model achieved an impressive 90.9% accuracy on test data, demonstrating resilience to noise, while the Random Forest model surpassed success criteria with 87.8% accuracy. The research highlights the potential of AI to enhance personal fitness training, making it safer and reducing injury risks, supporting the continued growth of AI-based solutions in the fitness industry.

AB - This study explores AI-driven form analysis in strength training, leveraging well-known techniques such as Convolutional Neural Networks (CNN) and Random Forest. The CNN model achieved an impressive 90.9% accuracy on test data, demonstrating resilience to noise, while the Random Forest model surpassed success criteria with 87.8% accuracy. The research highlights the potential of AI to enhance personal fitness training, making it safer and reducing injury risks, supporting the continued growth of AI-based solutions in the fitness industry.

KW - Artificial Intelligence

KW - Computer Vision

KW - Fitness

KW - Health

M3 - Paper

T2 - International Conference on Artificial Intelligence for Healthcare 2024

Y2 - 8 January 2024 through 9 January 2024

ER -

AI-driven form analysis: Personalised strength-training feedback via computer vision techniques

Abstract

Conference

Keywords

Fingerprint

Cite this