Prayer Activity Monitoring and Recognition Using Acceleration Features with Mobile Phone

Abstract

The increasing inclusion of sensors in mobile smartphones opens up new avenues for data mining applications for activity recognition. The objective is to detect the actions of one or more users from a series of observations regarding users’ body movements. In this project, we introduces a mobile-based application to monitor and recognise prayer activities (i.e. standing, bowing, prostration and sitting) using mobile phone acceleration features to determine the correctness of the prayer (i.e. the completeness and order of activities). The accelerometer data were collected for six prayers, totalling 118 samples, representing four main prayer activities. The collected data were used to train and test supervised machine learning algorithms to extract and recognise the prayer activities. Our experiments show that the prayer stages can be extracted and recognised accurately using machine learning algorithms. The WEKA machine learning toolkit was used to test classifiers using the features extracted from the accelerometer data. Three different classifiers were tested: Naive Bayes, IB1 Algorithm and the J48 Decision Trees and their accuracy exceeded 90 %.