Automated Posture Checker (APC)

The Automated Posture Checker is a system that monitors the posture of a user. Before the user begins work at a desk, the user would set up the APC by placing the camera at a position where it gets a clear side-profile full-body shot of himself/herself in a sitting position. The APC can then notify the user to correct their posture if the user slips into what is considered “bad” posture during their work session. Additionally, the APC can also remind the user to periodically change posture or move around to stretch. If the APC has determined that the user has been staying stiff and inactive for a long period of time, it will notify the user to change posture and/or move around and stretch. At the end of the work session, the user will stop the APC and get feedback on the overall properties and statistics of their posture and movement throughout their entire work session such as how long the session was and how long they had poor posture.

System Features:

• Create an account
• Create new sessions with customizable parameters and view previous sessions
• Receive alerts of when they should get up and stretch or have bad posture including an image of their bad posture and a constructive comment
• View statistics about their completed session through android application
• Detect posture with ML model and perform angle-based analysis to determine body posture

Notably Challenging features:

• Getting the camera to work
• Balancing ML accuracy with server performance

Authors: Justin Dar, Ken Liang, Dylan Pither, Jefferson Zhai

Roles in Project:

• Jefferson: All Hardware components
• Justin: Cloud component (ML)
• Ken: Cloud component (General)
• Dylan: All User components