Fitness ON - Game ON
Coach: Marius Lumpan
Team: Stefan Paici 10 yo, Selene Stoiovici 11 yo, Darius Cimpian 12 yo, Andrei Carsian 14 yo, Andrada Bratescu 14 yo, Iulia Lumpan 15 yo
Awards: Champions at FLL RePlay challange Bucharest
Contact: inventikasensoron@gmail.com
One of the biggest problems of the modern world is the high rate of people with back problems, especially with issues regarding the cervical and lumbar spine. The main cause of these problems is the incorrect posture we take when we sit for hours on end in front of our PCs. Another cause are the heavy backpacks we have to go to school with every day.
At this point we thought about an app that would assist the users in doing the exercises correctly. But how are we going to do this?
We wanted that, when running the application on a TV or a laptop, a silhouette would appear on the screen in which the user would have to fit in. With the help of the camera, you can see if the exercise is done correctly, and if the posture needs to be corrected, different audible signals will be heard. These sessions can be recorded, and later, the specialist can watch them to modify the series of exercises, based on the patient's progress.
For motivating the app users to exercise, we use the psychology behind video games. If the patient does the certain exercises, they get points that they can spend in other games (in-game currency). This way, dopamine is secreted and the self-esteem of the player rises. They can later spend the points on things they’re interested in, and with that in mind, they can stay motivated to continue using the sessions with our app, Fitness ON – Game ON
Because last summer we learned the basics of programming in Python and because Alex and Andrei had already been using the OpenCV module (the library used for processing video files) on a personal project, we started looking for solutions.
At first, I managed to adapt a program found on Github for face detection. We later moved on to body detection, but we found solutions that were 3 or 4 years old and didn’t even work properly. While we were trying to fix the errors we reached out to Mircea Cimpian, expert in programming and data bases.
That’s how we discovered TensorFlow. Created by Google Brain, TensorFlow is an open-source platform for calculations and high-level machine learning. From TensorFlow we use the MobileNet library, developed at Cornel University. So, our program imports this library, then it builds 19 points in 3D (on the x, y and z axes) and it draws the “skeleton” of the user. It then validates the different movements of every point that was set..
We talked about our project with team Inventika Storm and with team Gophistii, we presented it at the National College “Gheorghe Lazar”, National College “Gheorghe Sincai” and the Theoretical Highschool “Eugen Lovinescu” and, of course, we talked about it with our family and friends and received a positive feedback. Right now, we believe that our app Fitness On – Game ON could be used for any kind of exercise, not only for the ones regarding back problems. We would also like to collaborate with the other two FLL teams, with Storm because they have a complementary solution for blind people and with Gophistii because they bring the solution of a larger series of exercises in much more diverse environments.