The objective is to implement a system for human activity detection using monitorized data by means of smartphone sensors. First, we have to preprocess the signals applying recommended techniques from the paper Activity Recognition Using Inertial Sensing for Healthcare, Wellbeing and Sports Applications: A Survey (original_work.pdf). Then, using machine learning algorithms with hyper-parameter tuning, we make the predictions.
The data is composed by sequences of sensor data corresponding to different persons performing a set of activities in seminaturalistic conditions. It corresponds with AR_database.mat.
The implementation with all explanations and details corresponds with the notebook human_activity_sensor.ipynb.