Optimization of Sensor Placement Combinations and Classification Thresholds for the Accelerometer-Based Activity Recognition

Kwon, Yuri; Ho, Yeji; Heo, Jae-Hoon; Jeon, Hyeong Min; Kim, Jiwon; Eom, Gwang-Moon

doi:10.1166/jmihi.2018.2232

Cited by 3 publications

(5 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, these data were collected by using a structured format, for short durations (<10 min), and were, therefore, unlikely to capture the posture variations present in unobserved free-living. This variation would reduce the accuracy of threshold-based classifiers, and this is supported by the findings of Kwon et al [11], who found poorer classification accuracy (66.7%, SD = 6.0%) when using the same classifier on free-living data. Archer et al [12] attempted to classify postures using both accelerometers and gyroscopes along with a decision tree and k-nearest neighbour classifier, achieving 95% sensitivity.…”

Section: Introductionmentioning

confidence: 54%

“…The performance achieved by these models highlights the advantage of developing a classifier using machine learning methods as opposed to threshold-based classifiers. Previous research has shown that the accuracy of threshold-based classifiers is reduced by up to 40% when tested on free-living data compared to previously presented results on laboratory data [11]. This has been attributed to the variation in the execution of different postures typically found in free-living conditions, as opposed to the strict and considered movements performed in a laboratory.…”

Section: Discussionmentioning

confidence: 74%

“…Classifying amputee postures using a single shank worn accelerometer has only been explored by using threshold-based classifiers [6,11]. Redfield et al [6] proposed a binary decision tree classifier based on accelerometer signal magnitude area for classifying postures, movement, and wear, achieving high classification accuracy (96.6%, SD = 3%).…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A Machine Learning Classification Model for Monitoring the Daily Physical Behaviour of Lower-Limb Amputees

Griffiths

Diment

Granat

2021

Sensors

View full text Add to dashboard Cite

There are currently limited data on how prosthetic devices are used to support lower-limb prosthesis users in their free-living environment. Possessing the ability to monitor a patient’s physical behaviour while using these devices would enhance our understanding of the impact of different prosthetic products. The current approaches for monitoring human physical behaviour use a single thigh or wrist-worn accelerometer, but in a lower-limb amputee population, we have the unique opportunity to embed a device within the prosthesis, eliminating compliance issues. This study aimed to develop a model capable of accurately classifying postures (sitting, standing, stepping, and lying) by using data from a single shank-worn accelerometer. Free-living posture data were collected from 14 anatomically intact participants and one amputee over three days. A thigh worn activity monitor collected labelled posture data, while a shank worn accelerometer collected 3-axis acceleration data. Postures and the corresponding shank accelerations were extracted in window lengths of 5–180 s and used to train several machine learning classifiers which were assessed by using stratified cross-validation. A random forest classifier with a 15 s window length provided the highest classification accuracy of 93% weighted average F-score and between 88 and 98% classification accuracy across all four posture classes, which is the best performance achieved to date with a shank-worn device. The results of this study show that data from a single shank-worn accelerometer with a machine learning classification model can be used to accurately identify postures that make up an individual’s daily physical behaviour. This opens up the possibility of embedding an accelerometer-based activity monitor into the shank component of a prosthesis to capture physical behaviour information in both above and below-knee amputees. The models and software used in this study have been made open source in order to overcome the current restrictions of applying activity monitoring methods to lower-limb prosthesis users.

show abstract

Section: Introductionmentioning

confidence: 54%

Section: Discussionmentioning

confidence: 74%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A Machine Learning Classification Model for Monitoring the Daily Physical Behaviour of Lower-Limb Amputees

Griffiths

Diment

Granat

2021

Sensors

View full text Add to dashboard Cite

show abstract

“…Kwon et al . studied the wearing position of the sensor in human motion recognition and set the wireless three‐axis accelerators at different positions. The experimental results showed that the wearing position with the highest recognition rate of single‐sensor was the wrist, the best combination of sensors was to wear them at arms and thighs, and the best combination of three sensors was to wear them at arms, thighs, and ankles.…”

Section: Introductionmentioning

confidence: 99%

“…Hsu et al [6] identified 10 common family activities and 11 sports activities using two inertial sensors and found that the recognition rates were 98.23 and 99.55% respectively, which verified the reliability of the method. Kwon et al [7] studied the wearing position of the sensor in human motion recognition and a Correspondence to: Lei Zhao. E-mail: zleizhaol@yeah.net Department of Physical Education, Tianjin University of Commerce, Tianjin, 300134, China set the wireless three-axis accelerators at different positions.…”

Section: Introductionmentioning

confidence: 99%

Detection and recognition of human body posture in motion based on sensor technology

Zhao

Chen

2020

IEEJ Transactions Elec Engng

View full text Add to dashboard Cite

The recognition of human motion posture is of great values in the field of sports. In this paper, inertial sensor technology is employed to recognize four postures including dribbling, passing, catching, and shooting in basketball. The data are collected by four nine‐axis inertial sensors worn on the arm. The time domain and frequency domain features are extracted after the process of smoothing and normalizing. Then 30 eigenvectors are obtained by dimensionality reduction of principal component analysis (PCA), and the support vector machine (SVM) method is used for posture recognition. The experimental results show that after PCA dimensionality reduction, the recognition accuracy of the features as the SVM input is significantly high. Compared with the recognition accuracy of back propagation neural network (BPNN) (85.4%), the average recognition accuracy of SVM is 96%, which verifies the reliability of the method proposed in this study. This research proves the effectiveness of sensor technology in basketball posture recognition, which can provide a reliable guidance for basketball training. © 2020 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

show abstract

Posture and Physical Activity Detection: Impact of Number of Sensors and Feature Type

Tang

John

Thapa-Chhetry

et al. 2020

Medicine &Amp; Science in Sports &Amp; Exercise

View full text Add to dashboard Cite

Studies using wearable sensors to measure posture, physical activity (PA), and sedentary behavior typically use a single sensor worn on the ankle, thigh, wrist, or hip. Although the use of single sensors may be convenient, using multiple sensors is becoming more practical as sensors miniaturize. Purpose We evaluated the effect of single-site versus multisite motion sensing at seven body locations (both ankles, wrists, hips, and dominant thigh) on the detection of physical behavior recognition using a machine learning algorithm. We also explored the effect of using orientation versus orientation-invariant features on performance. Methods Performance (F1 score) of PA and posture recognition was evaluated using leave-one-subject-out cross-validation on a 42-participant data set containing 22 physical activities with three postures (lying, sitting, and upright). Results Posture and PA recognition models using two sensors had higher F1 scores (posture, 0.89 ± 0.06; PA, 0.53 ± 0.08) than did models using a single sensor (posture, 0.78 ± 0.11; PA, 0.43 ± 0.03). Models using two nonwrist sensors for posture recognition (F1 score, 0.93 ± 0.03) outperformed two-sensor models including one or two wrist sensors (F1 score, 0.85 ± 0.06). However, two-sensor models for PA recognition with at least one wrist sensor (F1 score, 0.60 ± 0.05) outperformed other two-sensor models (F1 score, 0.47 ± 0.02). Both posture and PA recognition F1 scores improved with more sensors (up to seven; 0.99 for posture and 0.70 for PA), but with diminishing performance returns. Models performed best when including orientation-based features. Conclusions Researchers measuring posture should consider multisite sensing using at least two nonwrist sensors, and researchers measuring PA should consider multisite sensing using at least one wrist sensor and one nonwrist sensor. Including orientation-based features improved both posture and PA recognition.

show abstract

Optimization of Sensor Placement Combinations and Classification Thresholds for the Accelerometer-Based Activity Recognition

Cited by 3 publications

References 0 publications

A Machine Learning Classification Model for Monitoring the Daily Physical Behaviour of Lower-Limb Amputees

A Machine Learning Classification Model for Monitoring the Daily Physical Behaviour of Lower-Limb Amputees

Detection and recognition of human body posture in motion based on sensor technology

Posture and Physical Activity Detection: Impact of Number of Sensors and Feature Type

Contact Info

Product

Resources

About