Edge.fit is a virtual personal fitness trainer which uses Computer vision, IoT, Intelligent Edge and Cloud computing to help users achieve their fitness goals efficiently.
Overview: Introduction Video
That which is not measured is not improved. Although there are many apps to record and analyse gym workouts, they need user input. By connecting the equipments at the gym, we can democratise having a personal trainer.
By coming in contact with the gym equipment, the weight, seat and various other settings are auto adjusted as per the user's body and training plan. The equipment will also measure the work out details like reps, sets, weights and store in the cloud. By analysing the pattern, the app will also provide personlised training plan. On equipping the gym with cameras, user will also get recommendations to improve posture and prevent injuries As part of the hackathon, based on the skill sets of the team members, idea is to get at least 2 of the above to a POC stage.
- Install Yarn
- Go to the directory of the project and type "yarn watch"
- We have also added the Azure Smart Trainer Bot for the integration of the user interface. You can use Azure Bot by running
python GymTrainerBot.py
inside the directory of "AzureGymTrainer".
- Or else you can also use the GUI created for this project in the Microsoft Hackathon by opening the folder "edgehack-final-project"
- Install the required modules in the requirements.txt in the local or the conda virtual enviroment.
- Then you can direclty execute the GUI by typing
python main.py
which will run the video through the integration with the Azure Bot.- Or to run directly posture recognition part of the project you can use
python demo.py
file.
- Azure Cognitive Services
- Pyttsx3
- LUIS
- Socket.io
- Flask
- Python 3
- PyTorch
- Posnet Model
We have used the Azure Congnitive Services for the Speech to Text and LUIS integration for the language understanding and the intent analyzation. GUI is created using the flask and UI bindings are done using socket.io asynchronous queues.
YouTube Link: Prototype Link