A Classification based task of predicting human activity (Sitting, standing, walking etc) based on the raw sensor data from the Accelerometer and Gyroscope.
-
Types of MLPs tried are,
- Used basic multi layered LSTM since it's a time series data of overlapping windows.
- Tried basic 1D Convolutions.
- Implemented a divide on conquer based technique of dynamic and static activity models as provided in reference-1.
-
References: