Activity Recognition Using One Triaxial Accelerometer: A Neuro-fuzzy Classifier with Feature Reduction

Yang, Jhun-Ying; Chen, Yen-Ping; Lee, Gwo-Yun; Liou, Shun-Nan; Wang, Jeen-Shing

doi:10.1007/978-3-540-74873-1_47

Cited by 40 publications

(30 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, the raw motion signature for the wrist while walking holding a heavy bag, walking with a full cup, or walking with a mobile phone held to the head may differ. Several recent studies explored the problem of detecting activities from wrist-worn sensors (8, 14, 25, 31, 32). As described in the discussion section, the prior studies have limitations due to the amount of data tested, the complexity of the activities tested, or the validation approach used when reporting results.…”

Section: Introductionmentioning

confidence: 99%

Activity Recognition Using a Single Accelerometer Placed at the Wrist or Ankle

Mannini

Intille

Rosenberger

et al. 2013

Medicine &Amp; Science in Sports &Amp; Exercise

343

305

View full text Add to dashboard Cite

PURPOSE Large physical activity surveillance projects such as the UK Biobank and NHANES are using wrist-worn accelerometer-based activity monitors that collect raw data. The goal is to increase wear time by asking subjects to wear the monitors on the wrist instead of the hip, and then to use information in the raw signal to improve activity type and intensity estimation. The purpose of this work is obtaining an algorithm to process wrist and ankle raw data and classify behavior into four broad activity classes: ambulation, cycling, sedentary and other. METHODS Participants (N = 33) wearing accelerometers on the wrist and ankle performed 26 daily activities. The accelerometer data were collected, cleaned, and preprocessed to extract features that characterize 2 s, 4 s, and 12.8 s data windows. Feature vectors encoding information about frequency and intensity of motion extracted from analysis of the raw signal were used with a support vector machine classifier to identify a subject’s activity. Results were compared with categories classified by a human observer. Algorithms were validated using a leave-one-subject-out strategy. The computational complexity of each processing step was also evaluated. RESULTS With 12.8 s windows, the proposed strategy showed high classification accuracies for ankle data (95.0%) that decreased to 84.7% for wrist data. Shorter (4 s) windows only minimally decreased performances of the algorithm on the wrist to 84.2%. CONCLUSIONS A classification algorithm using 13 features shows good classification into the four classes given the complexity of the activities in the original dataset. The algorithm is computationally-efficient and could be implemented in real-time on mobile devices with only 4 s latency.

show abstract

Section: Introductionmentioning

confidence: 99%

Activity Recognition Using a Single Accelerometer Placed at the Wrist or Ankle

Mannini

Intille

Rosenberger

et al. 2013

Medicine &Amp; Science in Sports &Amp; Exercise

343

305

View full text Add to dashboard Cite

show abstract

“…Many types of classifiers were evaluated in literature, including decision trees [13], k-nearest neighbor [14], support vector machines [15], naïve Bayesian [16], Bayesian net [17], artificial neural networks [18], linear discriminant analysis [19], and others [7]. To build or generate a classifier, offline training needs to be done.…”

Section: B Mode Of Motion Recognitionmentioning

confidence: 99%

Using portable device sensors to recognize height changing modes of motion

Elhoushi

Georgy

Wahdan

et al. 2014

2014 IEEE International Instrumentation and Measurement Technology Conference (I2MTC) Proceedings

View full text Add to dashboard Cite

In portable navigation, the need to recognize the motion mode of the user, is useful -if not necessary -to improve the positioning estimation. This paper explains the use of sensors in a portable device, such as a smartphone, to recognize the user's mode of motion when a change in height is detected. The modes of motion detected are walking up or down stairs, taking the elevator, and standing or walking on an escalator. The portable device contains an accelerometer triad, gyroscope triad, a barometer, and occasionally a magnetometer triad. The solution is dependent on the sensors, and does not require satellite or wireless positioning. The height motion mode recognition module has been implemented in real-time on several brands of various consumer portable devices, including smartphones, tablets, smartwatches, and smart glasses.

show abstract

“…2) Sometimes Used Features: These features are used in some works and include skewness [7], [8], [9], kurtosis [7], [8], signal magnitude area (SMA) [10], [11], [12], root mean square (RMS) [2], [13], signal magnitude vector (SVM) [14], eccentricity [7], spectral entropy [15], [16], [17], mean absolute deviation (MAD) [2], [13], interquartile range [2], [13], and FFT components [15], [17]. …”

Section: A Features Definitionmentioning

confidence: 99%

Applying a neuro-fuzzy classifier for gesture-based control using a single wrist-mounted accelerometer

Helmi

2009

2009 IEEE International Symposium on Computational Intelligence in Robotics and Automation - (CIRA)

View full text Add to dashboard Cite

Gestures, due to their natural modality, can be normally used in human-computer interaction (HCI) domains such as robotics, design environments and handheld devices. In this paper, a single wrist-mounted triaxial accelerometer is used to collect the acceleration data generated by hand movements forming 36 different gestures. This study intends to find the gestures which are capable of controlling an appliance with a maximum accuracy. A neuro-fuzzy classifier is devised for gestures detection to improve the classification rate in relative terms. The neuro-fuzzy system also selects the best features which yield the highest rate of classification. It reduces the dimensionality of feature set in two phases; the first phase is before carrying out the classification and the second phase is after selecting the most suitable gestures. The feature selection process finally reduces the number of features from 120 to 19. Our neuro-fuzzy system detects 25 gestures that can be classified with an accuracy of 100%, which is the highest rate among other classifiers. So, since the gesture-based control is accurately performed, it can be a proper method for HCI applications.

show abstract

Activity Recognition Using One Triaxial Accelerometer: A Neuro-fuzzy Classifier with Feature Reduction

Cited by 40 publications

References 4 publications

Activity Recognition Using a Single Accelerometer Placed at the Wrist or Ankle

Activity Recognition Using a Single Accelerometer Placed at the Wrist or Ankle

Using portable device sensors to recognize height changing modes of motion

Applying a neuro-fuzzy classifier for gesture-based control using a single wrist-mounted accelerometer

Contact Info

Product

Resources

About