A sensor based on recognition activities using smartphone
Marthen Dangu Elu Beily, Maria D Badjowawo, Daud Obed Bekak, Sumartini Dana · 2016
Nowadays, Technological advances facilitates monitoring of human lives through a set of sensors embedded in a smartphone. We can utilize these sensors for health care, monitoring elderly health, sports expert and entertainment. Many previous investigations have been conducted to perform accelerometer data for off-line recognition, but in this work, we attempted to propose the use of an activity recognition of tennis player in real-time. The values of the 3-axis accelerometer sensor were tested by sending these values to the server in real-time. A prototype application was developed to show and evaluate the selected classification methods for the designated recognition tennis activities. The results indicated that the Support Vector Machines (SVM) classifier using one second with window size of 20 samples obtained the highest accuracy of 96.25%. To measure the actual classification accuracy, the 10 fold cross validation was performed on data set using machine learning algorithm in Weka software.