A Feature Selection and Classification Method for Activity Recognition Based on an Inertial Sensing Unit

Abstract

The purpose of activity recognition is to identify activities through a series of observations of the experimenter’s behavior and the environmental conditions. In this study, through feature selection algorithms, we researched the effects of a large number of features on human activity recognition (HAR) assisted by an inertial measurement unit (IMU), and applied them to smartphones of the future. In the research process, we considered 585 features (calculated from tri-axial accelerometer and tri-axial gyroscope data). We comprehensively analyzed the features of signals and classification methods. Three feature selection algorithms were considered, and the combination effect between the features was used to select a feature set with a significant effect on the classification of the activity, which reduced the complexity of the classifier and improved the classification accuracy. We used five classification methods (support vector machine [SVM], decision tree, linear regression, Gaussian process, and threshold selection) to verify the classification accuracy. The activity recognition method we proposed could recognize six basic activities (BAs) (standing, going upstairs, going downstairs, walking, lying, and sitting) and postural transitions (PTs) (stand-to-sit, sit-to-stand, stand-to-lie, lie-to-stand, sit-to-lie, and lie-to-sit), with an average accuracy of 96.4%.

Supervised Machine Learning Applied to Wearable Sensor Data Can Accurately Classify Functional Fitness Exercises Within a Continuous Workout

“…these preliminary findings support the use of feature reduction in the pipeline of data processing, a more in-depth analysis of feature selection and outcomes derived thereof is advisable, especially for 1-sensor solutions with lower-end technology (Fan et al, 2019). Supervised machine learning appeared a suitable tool for the automatic classification of different functional fitness exercises.…”

“…Further analysis of feature selection suggested that the most informative characteristics of the dataset were mainly related to time domain (i.e., kurtosis and skewness). Although these preliminary findings support the use of feature reduction in the pipeline of data processing, a more in-depth analysis of feature selection and outcomes derived thereof is advisable, especially for 1-sensor solutions with lower-end technology (Fan et al, 2019 ).…”

Performance Comparisson Human Activity Recognition Using Simple Linear Method

“…Fan et. al [14] (98%) which used the same public HAR dataset and used SVM with the unknown kernel, but they have more features than as written by Anguita et.al. [16] that have 561 features, in [14] shown that used 585 features.…”

“…There are several methods to perform activity recognition: using multimedia (video), wearable devices (smart phone, smart watch) and ambient sensing [1] [11]. Several approaches were recognizing activities from one [12] [13] [14] [15] [16] or more sensor [10][17] [18] placement at the human body. Another approach based on vision was used in previous research but with some limitations, such an environmental restriction [14].…”

“…Several approaches were recognizing activities from one [12] [13] [14] [15] [16] or more sensor [10][17] [18] placement at the human body. Another approach based on vision was used in previous research but with some limitations, such an environmental restriction [14]. The camera amount used as the sensor was created a dimension of image as data.…”