Real-time physical training recognition

In this project we will learn effective technics for sitting workouts to make your everyday routines with a computer healthier, then I will show how to recognise different types of physical mouvements with artificial intelligence software and how to count a number of each mouvement authomatically with a micro:bit.

• Sitting workouts

Staying in the same position in the chair for long periods can cause poor posture, back pain, and reduced flexibility. Sitting workouts can help to improve your posture, strengthen your core muscles, and increase flexibility, which can help to reduce the risk of musculoskeletal disorders. By taking regular breaks to perform these exercises, you can improve blood flow and oxygen delivery to the brain and other organs, which can help to improve energy levels and reduce stress.

Here we will use four types of sitting exercises:

1. Head turns left and right

2. Deep breathing with shoulders mouvements

3. Stretching left and right using arm mouvements

4. Stretching upwords using arm mouvements

• Artificial intelligence model

We will create an AI model that will recognise in a real time what kind of mouvements you are doing. To do it, we will create 4 classes on the platform for image recognition with Teachable machine and each class will represent a specific type of sitting exercise that we mentioned above, then we will train and save the model.

I made a short video to explain how to set your AI model with Teachable machine:

• Set-up with the micro:bit

To connect a micro:bit with our AI model and count a number of each mouvements we chose an online platform for coding Vittascience. There are a lot of different platforms on the site, but for our mission we will use a coding interface Adacraft which is meant for Scratch projects. Here we will create a combinations of blocks to count each time when our AI model recognise that we do a specific mouvement (one of 4 types that we mentioned) and a micro:bit will count how many times we did it, as soon as we reach 10 times the micro:bit will show on the screen a fulfillment sign with a number of the class.

Here is a more detailed image with the code:

Hope you enjoyed working on the project and it will help you to stay healthy!