Toward Physical Activity Diary: Motion Recognition Using Simple Acceleration Features with Mobile Phones

In this paper, we perform physical motion recognition using mobile phones with built-in accelerometer sensors. Sensor data processing and smoothing techniques are discussed first to reduce the special noise present in phone-collected ac-celerometer data. We explore orientation-independent fea-tures extracted from vertical and horizonal components in acceleration as well as magnitudes of acceleration for six common physical activities, such as sitting, standing, walk-ing, running, driving and bicycling. We find decision tree achieves the best performance among four commonly used static classifiers, while vertical and horizonal features have better recognition accuracy than magnitude features. Fur-thermore, a well-pruned decision tree with simple time do-main features and less over-fitting on the training data can provide a usable model for inferencing a physical activity diary, refined by a similarity match from 푘-means clustering results and smoothed by an HMM-based Viterbi algorithm. Categories and Subject Descriptors I.5.2 [Design Methodology]: Classifier design and evalu-ation, Feature evaluation and selection, Pattern analysi