Human Activity Recognition using Smartphone

Human Activity Recognition using Smartphone

Balu^*, S. Vishaal Rangaprasad¹, A.C. Ajay Kumar²

* Assistant professor, Dept. of Computer Science and Engineering, SCSVMV, Kanchipuram

¹ Student, Department of Information technology, SCSVMV, Kanchipuram

² Student, Department of Computer Science and Engineering, SCSVMV, Kanchipuram

ABSTRACT

Sensor data can be recorded remotely, e.g. by video, radar or other wireless methods. It contains data generated by the smartphone's accelerometer, gyroscope and other sensors to train supervised prediction models using machine learning techniques such as SVM, random forest and decision tree to build a model. This can be used to predict the type of movement performed by the person, which is classified into six categories: Walking, climbing stairs, descending stairs, sitting, standing and lying down.

This research shows that the knowledge of experts does not differ semantically, but that the information tends to be interrelated. The experiments conducted have shown that the size of words and the speed with which words are retrieved from memory vary among individuals with different background knowledge.

Keywords:

Decision tree, Naïve Bayes, Random Forest, Sensors, Support vector machine.