As a prototype of wearable device for posture detection, this infrastructure consists of 3 major components: an Raspberry Pi Zero W, an IMU sensor with Bluetooth 2.0, and LEDs to provide feedback. For data acquisition and training mode, monitor, keyboard and mouse can be connected to the Raspberry Pi Zero to facilitate user inputs.
We implemented a program for posture detection on Python 3. In the posture labeling mode, the program allows the physical therapist to label postures (good standing posture, sway-back, and sitting.) In posture detection mode, the program uses the parameters determined through machine learning algorithm to classify posture and display feedback in terminal.
Link to Project Git Hub Repository:
Posture Detection Porject
Data Processing and Classification