Better Physical Activity Classification using Smartphone Acceleration Sensor

Abstract

Obesity is becoming one of the serious problems for the health of worldwide population. Social interactions on mobile phones and computers via internet through social e-networks are one of the major causes of lack of physical activities. For the health specialist, it is important to track the record of physical activities of the obese or overweight patients to supervise weight loss control. In this study, acceleration sensor present in the smartphone is used to monitor the physical activity of the user. Physical activities including Walking, Jogging, Sitting, Standing, Walking upstairs and Walking downstairs are classified. Time domain features are extracted from the acceleration data recorded by smartphone during different physical activities. Time and space complexity of the whole framework is done by optimal feature subset selection and pruning of instances. Classification results of six physical activities are reported in this paper. Using simple time domain features, 99 % classification accuracy is achieved. Furthermore, attributes subset selection is used to remove the redundant features and to minimize the time complexity of the algorithm. A subset of 30 features produced more than 98 % classification accuracy for the six physical activities.

Recognizing stationary and locomotion activities using combinational of spectral analysis with statistical descriptors features

“…Since only body acceleration signal used to determine the activity, gravity acceleration signal need to be eliminated. The cut-off frequency 0.3Hz is sufficient to separate those signal obtained [18]. If the value of cut-off frequency too high, it might tend to remove the required information from the signal.…”

“…In this work, we proposed two different categories of features from statistical descriptors and spectral analysis of frequency response. According to authors [18], features from statistical descriptors suitable for determining stationary activities like standing from sitting and etc. However, other periodic activities such as running and walking not really suitable, since those kinds of activities produce different statistical measures.…”

“…The noise needs to be removed for each dimension separately; x-axis, y-axis and zaxis. In this work, 5 th order Butterworth low pass filter with 0.3Hz cut-off frequency has been applied to separate acceleration signal into gravity and body acceleration [18]. The raw signal in the time domain will be transformed in term of a frequency domain using Fast Fourier Transform before filtering process conducted.…”

“…Two different groups of features were applied in this work; features from statistical descriptors and several features from a spectral measurement of frequency analysis. Statistical descriptors are useful to determine the stationary activities such as sitting and standing from the rest [18]. However, the locomotion activities such as walking, jogging and running involve acceleration patterns in different directions.…”

Recognizing Complex Human Activities using Hybrid Feature Selections based on an Accelerometer Sensor

“…According to Hall and Smith (1998), the correlation based features selection (CFS) method is proven in reducing the number of features for real or artificial data. Arif et al (2014) carried out research on the accelerometer performance, using a kNN model and using CFS with a reduced scatter search feature selection. Other related work in feature selection also been done by Akhavian and Behzadan (2015).…”