GYM.02: your new "gym buddy"

Get moving with our intelligents gyms

It is not a suprise that a lot of people like to work out: they want to be in good shape, feel good, impress somebody or just have a "summer body" all year round. And, there is also a lot of people who would like to start working out but is just afraid or just doesn't know anything about routines, machines and good positions. That's why I tought that about a new version of gyms: The GYM.02, your new gym buddy.

The concept: What is the GYM.02? It is a new and improved version of gyms. Just imagine, a gym where you enter and are instantly recognized by the "gym" in itself. You would be greeted by the gym who would then procede to ask you what type of routine you want to do that day. Once a routine is chosen, it will be displayed on the mirrors for you to follow it.

So, the gym is going to track all your movements just like a "gym budy" and hype you up when you need it, tell you if you are executing the exercices in the right position and if not help you achieve the best position.

The concept in images

Don't just imagine it. Have a look at it

https://miro.medium.com/max/2625/1*HSCK-kBIWD0WBHkJgBS3Jw.jpeg

The result

https://cdn.glitch.com/0d425d99-fd58-4ec6-8eb6-a703ee486a1c%2Fthumbnails%2F1_HSCK-kBIWD0WBHkJgBS3Jw.jpg?1575819355847

The technology behind the scenes

How does it work?

Understand how to do this project

Facial Recognition

Smile, your are being watched

In order to create "intelligent gyms" that greet you and recognize you we should use a ml5.js toolin order to recognize a subjet, through a video. How to code it?

https://i.ytimg.com/vi/3EH1sS6wJs8/hqdefault.jpg

Gesture Recognition

For several persons at the same time

We would use TensorFlow.js and PoseNet: a machine learning that estimates Human pose in real time. Installing this technlogy in the mirros of the gyms would allow them to follow your movements to then procede to tell you if your position is the right one or not. How to code it?

https://cdn.glitch.com/0d425d99-fd58-4ec6-8eb6-a703ee486a1c%2Fdownload.jpg?v=1575823583619

Gesture Recognition

For one person

We could choose to detect the position of just one person at a time. We would do it with the ml5.js tools: PoseNet and TensorFlor.js. How to code it?

https://miro.medium.com/max/2856/0*JmfSG0YqReUOjyGH.

Detection system

Are you doing the right position?

After estimating your pose, we would use YOLO to compare your position to the right position. The technology would rate your pose and tell you how close you are to doing the perfect position.
How to code it?

Aenean elementum ligula

Do you have a question?