Left to Right: Michael Keller, Paul Morris, Mary Pengelley
Meeting Times: Thursdays at 9:30 AM
Room: Behavioral Sciences 405, Florida Atlantic University
Michael Keller -Undergraduate Researcher, Team Leader of GPA Team
Paul Morris– Graduate Researcher, PhD Candidate- Center for Complex Systems and Brain Sciences
Mary Pengelley, RPT, DPT, ATP–Director of Physical Therapy at Progressive Pediatric Therapy, Special Professional Advisor to 3D Camera Team
The Good Posture Application (GPA) came about when Michael Keller, at that time an FAU High school senior, discovered the Kinect camera at MPCR. Seeing the potential applications of the skeleton tracking that came with the camera, Michael met with his own physical therapist, Mary Pengelley, who suggested that he use it as a subconscious reminder of correcting his potentially harmful atypical posture that comes from his cerebral palsy. Michael realized he could take it a step further and that this development could be beneficial to everyone. The version of the GPA that exists now through the Kinect camera is a proof of concept. It can track up to 6 people at once, and give audio feedback based on whether or not a user is leaning over. We are currently starting to use Machine Learning to do the same skeleton tracking that the Kinect does. Once this is implemented, we will transition onto a low-cost camera, in order to make the GPA widely available for the general public.
The implementation of the GPA on a smartphone platform will make it widely available to the public. As for the usefulness of this application in pediatric therapy, we have seen from previous demonstrations that children who use this application are constantly engaged. If we can apply this application to their therapy sessions, more positive results will be seen due to the children’s greater motivation. Outside of the pediatric therapy realm, it is known that people who spend a great deal of time sitting during the day, especially in front of a computer or television, can develop long-term postural issues due to lack of activity. This application will also be useful for them; they will not need to disturb their work to correct these issues with this application running in the background. This is not the only thing such cameras that give biofeedback can be used for. A Convolutional Network can also be used to gather information about other animals as well, like turtles or horses.
Below you will find screenshots of a working prototype of the GPA. As you can see, it is returning a small green number for good posture and a large red number for bad posture. The closer the number is to 0 and the greener the number, the better the person’s posture is and vice versa for larger red numbers.
October 2017- Below you will find a promo of our latest version of the GPA. Observe that it now rates participants using a letter grade in addition to the coloration of the number that denotes the math behind the scenes. Observe also that it now can track up to 6 people at once, by the color of their skeleton, and that the program can also perform face detection. Our next goal is to implement facial recognition, so that the program is able to distinguish between specific people, not just colors.
November 2017- Below you will find the latest stable version of the GPA with the ability to track up to 6 people at once by skeleton color as well as to track people without lighting.
Copyright © 2019 | Machine Perception & Cognitive Robotics Laboratory - Center for Complex Systems and Brain Sciences - Florida Atlantic University