Skeleton-Based Posture Classification to Promote Safer Walker-Assisted Gait in Older Adults

Carlos A. Cifuentes; Marcela M\'unera; Monica Sinha; Sergio D. Sierra M.

arxiv: 2605.00890 · v1 · submitted 2026-04-27 · 💻 cs.CV · cs.AI· cs.LG

Skeleton-Based Posture Classification to Promote Safer Walker-Assisted Gait in Older Adults

Sergio D. Sierra M. , Monica Sinha , Marcela M\'unera , Carlos A. Cifuentes This is my paper

Pith reviewed 2026-05-09 20:51 UTC · model grok-4.3

classification 💻 cs.CV cs.AIcs.LG

keywords posture classificationskeleton datasmart walkersfall preventionolder adultsXGBoostgeometric modelmachine learning

0 comments

The pith

Skeleton-based models using geometric and XGBoost achieve high accuracy in classifying postures for safer walker-assisted gait in older adults.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper tests multiple machine learning models to classify postures from skeleton data in older adults using walkers. The goal is to enable smart walkers to detect unsafe postures and help prevent falls, a leading cause of injury in the elderly. Geometric models reached 89.9 percent accuracy for eight postures, while XGBoost hit 99.24 percent training accuracy for seventeen postures and near-perfect scores in simpler binary tasks. Deep learning models also performed well in distinguishing standing from sitting or walker use. These results suggest that reliable posture detection can improve human-robot interaction and safety in assistive devices.

Core claim

The authors demonstrate that skeleton data can be used to classify a range of postures during walker-assisted walking, with a geometric approach achieving 89.9% accuracy across eight posture categories and XGBoost reaching 99.24% training accuracy for seventeen categories. Binary classifications such as detecting walker usage or standing versus sitting exceed 99% accuracy with XGBoost. The work positions this classification capability as a foundation for smart walkers that can intervene to promote safer gait and reduce fall risks among older adults.

What carries the argument

Skeleton-based posture classification using geometric feature extraction and XGBoost ensemble learning to identify safe and unsafe postures in real time for assistive walkers.

If this is right

Real-time posture monitoring could trigger alerts or adjustments in smart walkers to correct gait issues before falls occur.
High accuracy in distinguishing standing from sitting supports automated assistance features like seat activation.
Effective classification enhances the safety and usability of robotic walkers in clinical and home settings.
The approach opens possibilities for personalized feedback based on individual posture patterns.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

Deploying these models on embedded hardware in walkers could create fully autonomous safety systems without constant human oversight.
Combining posture classification with other sensor data like force or acceleration might improve overall fall prediction accuracy.
Extending the model to detect transitions between postures could provide even earlier warnings for potential instability.
Validation across different walker designs and user demographics would be needed to confirm broad applicability.

Load-bearing premise

That the posture classification accuracies achieved in the study's dataset will persist in uncontrolled, real-world environments with sensor noise and natural variability in movements.

What would settle it

Observing a significant drop in classification accuracy, such as below 80%, when the models are tested on new data collected from older adults using walkers in everyday home or community settings without controlled conditions.

Figures

Figures reproduced from arXiv: 2605.00890 by Carlos A. Cifuentes, Marcela M\'unera, Monica Sinha, Sergio D. Sierra M..

**Figure 2.** Figure 2: Outline of the image acquisition process for pose landmark detection and further posture classification. eXtreme Gradient Boosting (XGBoost), [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗

**Figure 3.** Figure 3: Comparison of F1 Scores for models predicting choice of walker, [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

read the original abstract

Falls among older adults are a significant public health concern, leading to severe injuries, loss of independence, and increased healthcare costs. This study evaluates the effectiveness of various models, including a Geometric approach, XGBoost, SVM, and several deep learning architectures, in classifying walker usage, standing vs. sitting, and posture for smart walkers used. Geometric and XGBoost were the top performers. XGBoost achieved near-perfect training accuracy in binary classification tasks, with 99.84% for walker choice and 99.69% for standing vs. sitting. For posture classification, Geometric approach attained 89.9% accuracy for 8 postures, and XGBoost obtained 99.24% during training for 17 postures. Deep learning models such as the 4-layer CNN and Encoder-Decoder CNN also demonstrated strong performance in binary classification, with accuracies above 98%. This study underscores the potential of machine learning to enhance human-robot interaction in smart walkers, particularly for fall prevention.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper applies standard classifiers to skeleton data from smart walkers and gets high numbers, but reports mostly training accuracy with almost no test-set or robustness details.

read the letter

The core takeaway is that this is a straightforward empirical application of existing models—Geometric baseline, XGBoost, SVM, and a couple of CNN variants—to classify 8 or 17 postures from walker-mounted skeleton data, aimed at fall prevention in older adults. Geometric hits 89.9% on the 8-posture task and XGBoost reaches 99.24% training accuracy on the 17-posture version, with similar near-perfect training figures for the binary walker-choice and stand-vs-sit tasks. That is the actual content: they collected some data, ran off-the-shelf tools, and showed the numbers work on that collection. No new algorithm, loss function, or theoretical framing appears. What the work does reasonably is lay out a practical comparison in a relevant assistive-robotics setting and flag which models look promising for real-time posture monitoring. The public-health angle is clear enough that someone building smart walkers might find the baseline useful. The soft spots sit right in the reporting. The abstract and stress-test note both flag that peak results are labeled as training accuracy, with no mention of held-out test performance, cross-validation folds, error bars, or any noise-injection checks. If the skeleton data already contains the variations that occur in uncontrolled walking, the numbers might hold; nothing in the write-up shows that. Hyperparameter choices for XGBoost and the CNNs are also free parameters that could be tuned to the specific dataset. This is not a load-bearing flaw in the idea, but it is a real gap in the evidence presented. The paper is for people working on human-robot interaction or assistive devices who need quick empirical pointers on posture classification rather than for methodologists looking for new techniques. It is coherent on its own terms and engages the relevant literature on ML for gait monitoring, so it clears the bar for a serious referee. I would send it out with explicit instructions to add test-set results, split details, and at least basic robustness checks before acceptance.

Referee Report

2 major / 2 minor

Summary. The manuscript evaluates several machine learning approaches, including a Geometric method, XGBoost, SVM, and deep learning architectures such as 4-layer CNN and Encoder-Decoder CNN, for classifying walker usage, standing versus sitting, and multi-class postures from skeleton data. The goal is to support safer gait in older adults using smart walkers for fall prevention. Geometric and XGBoost are identified as top performers, with Geometric achieving 89.9% accuracy on 8 postures and XGBoost reaching 99.84% and 99.69% training accuracy on binary tasks plus 99.24% training accuracy on 17 postures; deep learning models exceed 98% on binary classification.

Significance. If the reported performance generalizes beyond the training data, the work would have moderate significance for computer vision applications in assistive robotics. It demonstrates the feasibility of skeleton-based posture detection for real-time human-robot interaction in healthcare settings, with the multi-model comparison and focus on older-adult gait providing a practical contribution to fall-prevention technologies.

major comments (2)

[Abstract] Abstract: The 99.24% figure for 17-posture classification is explicitly labeled as training accuracy for XGBoost, yet no test-set accuracy, cross-validation procedure, data-split ratios, or error bars are provided. This directly undermines the central claim that the models (particularly Geometric at 89.9% for 8 postures and XGBoost) can promote safer walker-assisted gait, because training fit alone does not establish generalization to uncontrolled environments or sensor noise.
[Results] Results (or equivalent section reporting model performance): No quantitative baseline comparisons to prior skeleton-based posture classifiers or ablation studies on feature choices (e.g., joint angles versus raw coordinates) are included, making it impossible to determine whether the reported accuracies represent a meaningful advance rather than dataset-specific fitting.

minor comments (2)

[Abstract] Abstract: The sentence structure listing accuracies mixes binary and multi-class results without clear separation, which reduces readability; rephrasing to group binary versus posture tasks would improve clarity.
[Methods] The manuscript would benefit from an explicit statement of the total number of participants, skeleton extraction method (e.g., OpenPose version), and any preprocessing steps for noise reduction.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments on our manuscript. We address each major point below and describe the revisions we will undertake to improve clarity and rigor.

read point-by-point responses

Referee: [Abstract] Abstract: The 99.24% figure for 17-posture classification is explicitly labeled as training accuracy for XGBoost, yet no test-set accuracy, cross-validation procedure, data-split ratios, or error bars are provided. This directly undermines the central claim that the models (particularly Geometric at 89.9% for 8 postures and XGBoost) can promote safer walker-assisted gait, because training fit alone does not establish generalization to uncontrolled environments or sensor noise.

Authors: We agree that the abstract's focus on training accuracy for the 17-posture XGBoost result, without accompanying test metrics, risks overstating generalization. The full manuscript reports the Geometric method's 89.9% accuracy on the 8-posture task as overall performance on the collected dataset, but we acknowledge the need for explicit test-set evaluation. In the revision we will update the abstract and results to report held-out test accuracies, the cross-validation procedure used, train/test split ratios, and error bars across folds or runs. This will more accurately support the claims regarding applicability to fall-prevention scenarios. revision: yes
Referee: [Results] Results (or equivalent section reporting model performance): No quantitative baseline comparisons to prior skeleton-based posture classifiers or ablation studies on feature choices (e.g., joint angles versus raw coordinates) are included, making it impossible to determine whether the reported accuracies represent a meaningful advance rather than dataset-specific fitting.

Authors: We concur that the lack of explicit baselines and ablations makes it difficult to gauge the advance relative to existing skeleton-based posture work. We will add a dedicated comparison subsection that tabulates our results against representative prior methods on similar skeleton data. We will also include ablation experiments contrasting joint-angle features against raw coordinate inputs for both the Geometric and XGBoost pipelines, reporting the resulting accuracy differences. These additions will clarify the contribution of our modeling choices. revision: yes

Circularity Check

0 steps flagged

No significant circularity in standard supervised ML classification pipeline

full rationale

The paper reports empirical performance of models (Geometric, XGBoost, SVM, CNNs) trained on collected skeleton data to classify walker usage, standing/sitting, and postures. Accuracies such as 89.9% for 8 postures and 99.24% training accuracy for 17 postures are direct outputs of model fitting on external data. No equations, derivations, or first-principles results are claimed that reduce to inputs by construction. No self-citations, ansatzes, or uniqueness theorems are load-bearing. This is a self-contained data-driven study.

Axiom & Free-Parameter Ledger

2 free parameters · 2 axioms · 0 invented entities

The central claim rests on standard machine-learning assumptions about data quality and model generalization rather than new theoretical constructs or invented entities.

free parameters (2)

XGBoost and CNN hyperparameters
Tuned to achieve the reported training accuracies on the posture dataset
Choice of 8- and 17-posture label sets
Selected post-hoc to demonstrate performance

axioms (2)

domain assumption Skeleton keypoints from sensors accurately represent human posture and gait dynamics
Invoked throughout as the basis for all classification inputs
domain assumption The experimental dataset reflects typical walker use by older adults in relevant settings
Required for the reported accuracies to imply real-world utility

pith-pipeline@v0.9.0 · 5486 in / 1400 out tokens · 52786 ms · 2026-05-09T20:51:47.145342+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

23 extracted references · 2 canonical work pages

[1]

Nations,Ageing, 2024

U. Nations,Ageing, 2024. [Online]. Available:https://www. un.org/en/global-issues/ageing

2024
[2]

Observational evidence of the association between physical and psychological determinants of aging with cognition in older adults,

V . F. Martins et al., “Observational evidence of the association between physical and psychological determinants of aging with cognition in older adults,”Scientific Reports, vol. 14, p. 12 574, 1 May 2024,ISSN: 2045-2322

2024
[3]

N. I. on Aging,Falls and fractures in older adults: Causes and prevention, 2022. [Online]. Available:https : / / www . nia . nih . gov / health / falls - and - falls - prevention / falls- and- fractures- older- adults- causes- and- prevention

2022
[4]

Factors influencing the use of walking aids by frail elderly people in senior day care centers,

Y . Sakano et al., “Factors influencing the use of walking aids by frail elderly people in senior day care centers,”Healthcare, vol. 11, p. 858, 6 Mar. 2023,ISSN: 2227-9032

2023
[5]

Mobility device use in older adults and incidence of falls and worry about falling: Findings from the 2011–2012 national health and aging trends study,

N. M. Gell et al., “Mobility device use in older adults and incidence of falls and worry about falling: Findings from the 2011–2012 national health and aging trends study,”Journal of the American Geriatrics Society, vol. 63, pp. 853–859, 5 May 2015,ISSN: 0002- 8614

2011
[6]

Perspectives on use of mobility aids in a diverse population of seniors: Implications for intervention,

L. Resnik et al., “Perspectives on use of mobility aids in a diverse population of seniors: Implications for intervention,”Disability and Health Journal, vol. 2, pp. 77–85, 2 Apr. 2009,ISSN: 19366574

2009
[7]

Wearable devices for ergonomics: A systematic literature review,

E. Stefana et al., “Wearable devices for ergonomics: A systematic literature review,”Sensors, vol. 21, p. 777, 3 Jan. 2021,ISSN: 1424- 8220

2021
[8]

Comparative analysis of skeleton-based human pose estimation,

J.-L. Chung et al., “Comparative analysis of skeleton-based human pose estimation,”Future Internet, vol. 14, p. 380, 12 Dec. 2022, ISSN: 1999-5903

2022
[9]

Sensor-based wearable systems for monitoring human motion and posture: A review,

X. Huang et al., “Sensor-based wearable systems for monitoring human motion and posture: A review,”Sensors, vol. 23, p. 9047, 22 Nov. 2023,ISSN: 1424-8220

2023
[10]

Human posture detection using image augmentation and hyperparameter-optimized transfer learning algo- rithms,

R. O. Ogundokun et al., “Human posture detection using image augmentation and hyperparameter-optimized transfer learning algo- rithms,”Applied Sciences, vol. 12, p. 10 156, 19 Oct. 2022,ISSN: 2076-3417

2022
[11]

A comprehensive survey on 2d multi-person pose estimation methods,

C. Wang et al., “A comprehensive survey on 2d multi-person pose estimation methods,”Engineering Applications of Artificial Intelligence, vol. 102, p. 104 260, Jun. 2021,ISSN: 09521976

2021
[12]

Deep learning of human posture image classifi- cation using convolutional neural networks,

A. R. Rababaah, “Deep learning of human posture image classifi- cation using convolutional neural networks,”International Journal of Computing Science and Mathematics, vol. 15, p. 273, 3 2022, ISSN: 1752-5055

2022
[13]

Socially assistive walker for daily living assistance in older adults,

S. D. Sierra M. et al., “Socially assistive walker for daily living assistance in older adults,”Frontiers in Robotics and AI, vol. 11, Aug. 2024,ISSN: 2296-9144. [Online]. Available:https://www. frontiersin.org/articles/10.3389/frobt.2024. 1401663/full

work page doi:10.3389/frobt.2024 2024
[14]

[Online]

Google Research,Pose landmark detection guide, 2024. [Online]. Available:https://ai.google.dev/edge/mediapipe/ solutions/vision/pose_landmarker/index#models

2024
[15]

Improved cnn based on batch normal- ization and adam optimizer,

R. O. Ogundokun et al., “Improved cnn based on batch normal- ization and adam optimizer,” inProceedings of the International Conference on Computational Science and Its Applications, Malaga, Spain: Springer, 2022, pp. 593–604

2022
[16]

An encoder-decoder based convolution neural network (cnn) for future advanced driver assistance system (adas),

R. Yasrab et al., “An encoder-decoder based convolution neural network (cnn) for future advanced driver assistance system (adas),” Applied Sciences, vol. 7, no. 4, p. 312, 2017. [Online]. Available: https://doi.org/10.3390/app7040312

work page doi:10.3390/app7040312 2017
[17]

Optimal image characterization for in-bed posture classification by using svm algorithm,

C. A. Rivera-Romero et al., “Optimal image characterization for in-bed posture classification by using svm algorithm,”Big Data and Cognitive Computing, vol. 8, no. 2, 2024,ISSN: 2504-2289. [Online]. Available:https://www.mdpi.com/2504-2289/ 8/2/13

2024
[18]

A comparative analysis of gradient boosting algorithms,

C. Bent ´ejac et al., “A comparative analysis of gradient boosting algorithms,”Artificial Intelligence Review, vol. 54, no. 3, pp. 1937– 1967, 2021. [Online]. Available:https : / / doi . org / 10 . 1007/s10462-020-09896-5

1937
[19]

Optimal image characterization for in- bed posture classification by using svm algorithm,

C. A. Rivera-Romero et al., “Optimal image characterization for in- bed posture classification by using svm algorithm,”Big Data and Cognitive Computing, vol. 8, p. 13, 2 Jan. 2024,ISSN: 2504-2289

2024
[20]

In-bed posture classification using deep neural network,

L. Stern et al., “In-bed posture classification using deep neural network,”Sensors, vol. 23, p. 2430, 5 Feb. 2023,ISSN: 1424-8220

2023
[21]

Accuracy of gait and posture classification using movement sensors in individuals with mobility impairment after stroke,

J. Pohl et al., “Accuracy of gait and posture classification using movement sensors in individuals with mobility impairment after stroke,”Frontiers in Physiology, vol. 13, Sep. 2022,ISSN: 1664- 042X

2022
[22]

Pressure-based posture classification methods and algorithms: A systematic review,

L. Fonseca et al., “Pressure-based posture classification methods and algorithms: A systematic review,”Computers, vol. 12, p. 104, 5 May 2023,ISSN: 2073-431X

2023
[23]

Classification of static postures with wearable sensors mounted on loose clothing,

U. Jayasinghe et al., “Classification of static postures with wearable sensors mounted on loose clothing,”Scientific Reports, vol. 13, p. 131, 1 Jan. 2023,ISSN: 2045-2322

2023

[1] [1]

Nations,Ageing, 2024

U. Nations,Ageing, 2024. [Online]. Available:https://www. un.org/en/global-issues/ageing

2024

[2] [2]

Observational evidence of the association between physical and psychological determinants of aging with cognition in older adults,

V . F. Martins et al., “Observational evidence of the association between physical and psychological determinants of aging with cognition in older adults,”Scientific Reports, vol. 14, p. 12 574, 1 May 2024,ISSN: 2045-2322

2024

[3] [3]

N. I. on Aging,Falls and fractures in older adults: Causes and prevention, 2022. [Online]. Available:https : / / www . nia . nih . gov / health / falls - and - falls - prevention / falls- and- fractures- older- adults- causes- and- prevention

2022

[4] [4]

Factors influencing the use of walking aids by frail elderly people in senior day care centers,

Y . Sakano et al., “Factors influencing the use of walking aids by frail elderly people in senior day care centers,”Healthcare, vol. 11, p. 858, 6 Mar. 2023,ISSN: 2227-9032

2023

[5] [5]

Mobility device use in older adults and incidence of falls and worry about falling: Findings from the 2011–2012 national health and aging trends study,

N. M. Gell et al., “Mobility device use in older adults and incidence of falls and worry about falling: Findings from the 2011–2012 national health and aging trends study,”Journal of the American Geriatrics Society, vol. 63, pp. 853–859, 5 May 2015,ISSN: 0002- 8614

2011

[6] [6]

Perspectives on use of mobility aids in a diverse population of seniors: Implications for intervention,

L. Resnik et al., “Perspectives on use of mobility aids in a diverse population of seniors: Implications for intervention,”Disability and Health Journal, vol. 2, pp. 77–85, 2 Apr. 2009,ISSN: 19366574

2009

[7] [7]

Wearable devices for ergonomics: A systematic literature review,

E. Stefana et al., “Wearable devices for ergonomics: A systematic literature review,”Sensors, vol. 21, p. 777, 3 Jan. 2021,ISSN: 1424- 8220

2021

[8] [8]

Comparative analysis of skeleton-based human pose estimation,

J.-L. Chung et al., “Comparative analysis of skeleton-based human pose estimation,”Future Internet, vol. 14, p. 380, 12 Dec. 2022, ISSN: 1999-5903

2022

[9] [9]

Sensor-based wearable systems for monitoring human motion and posture: A review,

X. Huang et al., “Sensor-based wearable systems for monitoring human motion and posture: A review,”Sensors, vol. 23, p. 9047, 22 Nov. 2023,ISSN: 1424-8220

2023

[10] [10]

Human posture detection using image augmentation and hyperparameter-optimized transfer learning algo- rithms,

R. O. Ogundokun et al., “Human posture detection using image augmentation and hyperparameter-optimized transfer learning algo- rithms,”Applied Sciences, vol. 12, p. 10 156, 19 Oct. 2022,ISSN: 2076-3417

2022

[11] [11]

A comprehensive survey on 2d multi-person pose estimation methods,

C. Wang et al., “A comprehensive survey on 2d multi-person pose estimation methods,”Engineering Applications of Artificial Intelligence, vol. 102, p. 104 260, Jun. 2021,ISSN: 09521976

2021

[12] [12]

Deep learning of human posture image classifi- cation using convolutional neural networks,

A. R. Rababaah, “Deep learning of human posture image classifi- cation using convolutional neural networks,”International Journal of Computing Science and Mathematics, vol. 15, p. 273, 3 2022, ISSN: 1752-5055

2022

[13] [13]

Socially assistive walker for daily living assistance in older adults,

S. D. Sierra M. et al., “Socially assistive walker for daily living assistance in older adults,”Frontiers in Robotics and AI, vol. 11, Aug. 2024,ISSN: 2296-9144. [Online]. Available:https://www. frontiersin.org/articles/10.3389/frobt.2024. 1401663/full

work page doi:10.3389/frobt.2024 2024

[14] [14]

[Online]

Google Research,Pose landmark detection guide, 2024. [Online]. Available:https://ai.google.dev/edge/mediapipe/ solutions/vision/pose_landmarker/index#models

2024

[15] [15]

Improved cnn based on batch normal- ization and adam optimizer,

R. O. Ogundokun et al., “Improved cnn based on batch normal- ization and adam optimizer,” inProceedings of the International Conference on Computational Science and Its Applications, Malaga, Spain: Springer, 2022, pp. 593–604

2022

[16] [16]

An encoder-decoder based convolution neural network (cnn) for future advanced driver assistance system (adas),

R. Yasrab et al., “An encoder-decoder based convolution neural network (cnn) for future advanced driver assistance system (adas),” Applied Sciences, vol. 7, no. 4, p. 312, 2017. [Online]. Available: https://doi.org/10.3390/app7040312

work page doi:10.3390/app7040312 2017

[17] [17]

Optimal image characterization for in-bed posture classification by using svm algorithm,

C. A. Rivera-Romero et al., “Optimal image characterization for in-bed posture classification by using svm algorithm,”Big Data and Cognitive Computing, vol. 8, no. 2, 2024,ISSN: 2504-2289. [Online]. Available:https://www.mdpi.com/2504-2289/ 8/2/13

2024

[18] [18]

A comparative analysis of gradient boosting algorithms,

C. Bent ´ejac et al., “A comparative analysis of gradient boosting algorithms,”Artificial Intelligence Review, vol. 54, no. 3, pp. 1937– 1967, 2021. [Online]. Available:https : / / doi . org / 10 . 1007/s10462-020-09896-5

1937

[19] [19]

Optimal image characterization for in- bed posture classification by using svm algorithm,

C. A. Rivera-Romero et al., “Optimal image characterization for in- bed posture classification by using svm algorithm,”Big Data and Cognitive Computing, vol. 8, p. 13, 2 Jan. 2024,ISSN: 2504-2289

2024

[20] [20]

In-bed posture classification using deep neural network,

L. Stern et al., “In-bed posture classification using deep neural network,”Sensors, vol. 23, p. 2430, 5 Feb. 2023,ISSN: 1424-8220

2023

[21] [21]

Accuracy of gait and posture classification using movement sensors in individuals with mobility impairment after stroke,

J. Pohl et al., “Accuracy of gait and posture classification using movement sensors in individuals with mobility impairment after stroke,”Frontiers in Physiology, vol. 13, Sep. 2022,ISSN: 1664- 042X

2022

[22] [22]

Pressure-based posture classification methods and algorithms: A systematic review,

L. Fonseca et al., “Pressure-based posture classification methods and algorithms: A systematic review,”Computers, vol. 12, p. 104, 5 May 2023,ISSN: 2073-431X

2023

[23] [23]

Classification of static postures with wearable sensors mounted on loose clothing,

U. Jayasinghe et al., “Classification of static postures with wearable sensors mounted on loose clothing,”Scientific Reports, vol. 13, p. 131, 1 Jan. 2023,ISSN: 2045-2322

2023