Fitletics is our capstone project which is a framework that helps the users keep up with their fitness goals. This is done by a web component that helps with the real-time tracking of exercises as well as an Android app that recommends workouts based on the body type and current fitness level of the user which then provides analytics on the past sessions
This is the website module that handles scanning the user's body to determine the body type along with having them tracked in real-time using the webcam. The body analysis is handled by a CNN trained on the SOMASET from Kaggle which had about 100k images along with the market-1505 dataset with a further 1.5k images. For the real-time rep counting, we implemented Google's PoseNet which is lightweight, yet accurate for detecting the different poses while in a workout
Clone the repository
git clone https://github.com/ByteSizeTech/fitletics-web.git
├── build # HTML files and support dirs
│ ├── Images # Support images
│ ├── JsonData # Support json files for models
└── ...
├── src
│ ├── Models # Body analysis' & Supported exercise's models
│ ├── Scripts # JavaScript logic files and classes
└── └── styles # .css for page styling