Simultaneous exercise recognition and evaluation in prescribed routines: Approach to virtual coaches

Villa, Sara García de; Casillas-Pérez, David; Martín, Ana Jiménez; Domínguez, Juan Jesús García

doi:10.1016/j.eswa.2022.116990

Cited by 13 publications

(12 citation statements)

References 204 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Table 5 contains the recent works relevant to our study, including exercise recognition (ER) and exercise assessment (EA). For the exercise assessment, which is the primary purpose of our study, we found two vision-based studies: [ 21 , 44 ]. Both studies provided roughly similar performance to our proposed model, whereas our dataset consisted of more types of exercises.…”

Section: Resultsmentioning

confidence: 99%

Design and Validation of Vision-Based Exercise Biofeedback for Tele-Rehabilitation

Khanghah

Fernie

Fekr

2023

Sensors

View full text Add to dashboard Cite

Tele-rehabilitation has the potential to considerably change the way patients are monitored from their homes during the care process, by providing equitable access without the need to travel to rehab centers or shoulder the high cost of personal in-home services. Developing a tele-rehab platform with the capability of automating exercise guidance is likely to have a significant impact on rehabilitation outcomes. In this paper, a new vision-based biofeedback system is designed and validated to identify the quality of performed exercises. This new system will help patients to refine their movements to get the most out of their plan of care. An open dataset was used, which consisted of data from 30 participants performing nine different exercises. Each exercise was labeled as “Correctly” or “Incorrectly” executed by five clinicians. We used a pre-trained 3D Convolution Neural Network (3D-CNN) to design our biofeedback system. The proposed system achieved average accuracy values of 90.57% ± 9.17% and 83.78% ± 7.63% using 10-Fold and Leave-One-Subject-Out (LOSO) cross validation, respectively. In addition, we obtained average F1-scores of 71.78% ± 5.68% using 10-Fold and 60.64% ± 21.3% using LOSO validation. The proposed 3D-CNN was able to classify the rehabilitation videos and feedback on the quality of exercises to help users modify their movement patterns.

show abstract

Section: Resultsmentioning

confidence: 99%

Design and Validation of Vision-Based Exercise Biofeedback for Tele-Rehabilitation

Khanghah

Fernie

Fekr

2023

Sensors

View full text Add to dashboard Cite

show abstract

“…Recent research has employed data-driven methodologies to estimate various biomechanical metrics and support exercise routines. The approach of García-de-Villa et al combined exercise recognition and evaluation as a single task, utilizing data from four IMUs and achieving accuracies between 88.4% and 91.4% [28]. Despite the excellent performance, we avoided the use of a sensor system due to cost and simplicity considerations.…”

Section: Discussionmentioning

confidence: 99%

Towards Automating Personal Exercise Assessment and Guidance with Affordable Mobile Technology

Sideridou,

Kouidi,

Hatzitaki

et al. 2024

Sensors

View full text Add to dashboard Cite

Physical activity (PA) offers many benefits for human health. However, beginners often feel discouraged when introduced to basic exercise routines. Due to lack of experience and personal guidance, they might abandon efforts or experience musculoskeletal injuries. Additionally, due to phenomena such as pandemics and limited access to supervised exercise spaces, especially for the elderly, the need to develop personalized systems has become apparent. In this work, we develop a monitored physical exercise system that offers real-time guidance and recommendations during exercise, designed to assist users in their home environment. For this purpose, we used posture estimation interfaces that recognize body movement using a computer or smartphone camera. The chosen pose estimation model was BlazePose. Machine learning and signal processing techniques were used to identify the exercise currently being performed. The performances of three machine learning classifiers were evaluated for the exercise recognition task, achieving test-set accuracy between 94.76% and 100%. The research methodology included kinematic analysis (KA) of five selected exercises and statistical studies on performance and range of motion (ROM), which enabled the identification of deviations from the expected exercise execution to support guidance. To this end, data was collected from 57 volunteers, contributing to a comprehensive understanding of exercise performance. By leveraging the capabilities of the BlazePose model, an interactive tool for patients is proposed that could support rehabilitation programs remotely.

show abstract

“…This database, called PHYTMO (from PHYsical Therapy MOnitoring) is created for its use in the development of novel algorithms for the evaluation of human motions, including the identification and assessment of a known set of prescribed exercises. The database includes enough data for developing ML-based algorithms, as the authors proved in their proposal for the exercises recognition and evaluation that uses part of the inertial data of PHYTMO 18 . For the development of robust and generalizable algorithms, it is required a large amount of annotated data and the subjects variability is also important.…”

Section: Background and Summarymentioning

confidence: 99%

“…Related to this topic, data can be used in ML-based studies for classification purposes, such as the identification of a set of exercises 48 . As a matter of fact, this database has been used for the simultaneous recognition and evaluation of exercises in physical routines 18 . Another possible use is the study of motion kinematics, focused on the raw inertial data or aiming to evaluate joint angles, an important parameter during these prescribed exercises 46 .…”

Section: Technical Validationmentioning

confidence: 99%

A database of physical therapy exercises with variability of execution collected by wearable sensors

2022

Self Cite

View full text Add to dashboard Cite

This document introduces the PHYTMO database, which contains data from physical therapies recorded with inertial sensors, including information from an optical reference system. PHYTMO includes the recording of 30 volunteers, aged between 20 and 70 years old. A total amount of 6 exercises and 3 gait variations were recorded. The volunteers performed two series with a minimum of 8 repetitions in each one. PHYTMO includes magneto-inertial data, together with a highly accurate location and orientation in the 3D space provided by the optical system. The files were stored in CSV format to ensure its usability. The aim of this dataset is the availability of data for two main purposes: the analysis of techniques for the identification and evaluation of exercises using inertial sensors and the validation of inertial sensor-based algorithms for human motion monitoring. Furthermore, the database stores enough data to apply Machine Learning-based algorithms. The participants’ age range is large enough to establish age-based metrics for the exercises evaluation or the study of differences in motions between different groups.

show abstract

Simultaneous exercise recognition and evaluation in prescribed routines: Approach to virtual coaches

Cited by 13 publications

References 204 publications

Design and Validation of Vision-Based Exercise Biofeedback for Tele-Rehabilitation

Design and Validation of Vision-Based Exercise Biofeedback for Tele-Rehabilitation

Towards Automating Personal Exercise Assessment and Guidance with Affordable Mobile Technology

A database of physical therapy exercises with variability of execution collected by wearable sensors

Contact Info

Product

Resources

About