At some point in their lives, around one fifth of the population will be in need of a physiotherapist, however waiting lists are long and heavily underserved.
To reduce the burden on physios, we've developed a system that helps deliver and track a user's progress with their exercise plan, utilising in-browser machine learning via the webcam to identify when users are performing said exercises and for how long.
This could have wider implications for exercise monitoring, motivation, and wellbeing:
- Exercise and stretch routines for the elderly
- Rehabilitation from injury and surgery
- Gamification of physio treatment plans
- Monitoring form
In addition, the same technologies can be used for various fitness activities, such as remote and automated classes for yoga, conditioning, pilates, stretches, and powerlifting.
Extensions could be adding on pose prediction and correction suggestions by modelling the body landmarks as a pose skeleton, perhaps using a Kinect or possibly still the webcam. The control system or UI could also be voice activated to interface while mid-exercise.
This was built with TensorflowJS with training and prediction done in-browser, so images never leave your machine for privacy, there is no reliance on a high broadband connection, and is scalable as there is significantly lower server load.
Built With
- javascript
- python
- tensorflow-js
Log in or sign up for Devpost to join the conversation.