rehab.ai

• 

  home page

• 

  exercise page

Categories this project falls in

Track: Innovating with AI Side Challenges: Sargent College Hack Best Domain Name BU Spark: Best College Life Hack

Inspiration

As exercise is the key to a healthy body, knowing how to correctly do each exercise is crucial to our physical and mental health. The lack of free instant feedbacks from rehab/fitness app has inspired us to create a free app that allows everyone from across the global to access this teaching. We believe our focus area is fundamental to reducing the workout errors by allowing people to better understand the specialty of workouts and improving the quality of care.

What it does

This projects educates people who are curious about the correctness of their exercising form. The user will input a video of them doing a workout and our website informs them if their form is correct or not.

How we built it

We created a pose-estimation model using the Python deep learning library tensorflow, created an API using flask(a micro web framework for Python) that parses through a video and predicts on each frame of the video determining whether a patient is correctly doing the exercise, and we used html, css, and javascript to create a mobile and desktop responsive website.

Challenges we ran into

Brainstorming: we took over 3 hours to brainstorm our idea

Addressing the possibility of transcribing errors: During our development process, several data cases derived transcribing errors interrupting the execution of the program.

Troubleshooting: There were countless occasions where we had to debug the code, especially for layered lists, in order to identify errors. We also had to learn new programming libraries and concepts to achieve the desired advanced effects for our goal.

Limited access to datasets and Privacy Concerns: We aim to establish collaboration with other colleges or even with physicians and medical institutes in the future to help others improve the workout aspect.

Accomplishments that we're proud of

First, we are proud of brainstorming and applying an innovative way of understanding the workout aspect for college students. We’re excited about the impact it would bring to help anyone to better understand their mistakes, facilitating an effective deep learning library for workouts. Lastly, we are proud of being able to create a working video upload page and ways of doing exercises correctly in a short time to prove the viability of our idea.

What we learned

By coming up with numerous ideas, we gained a better understanding of the mechanisms and problems with understanding the aspect of working out. The program development cycle also deepened our understanding of machine learning and web-development. Most importantly, we experienced the ins and outs of the hackathon, broadening our understanding of the rehab world, and learned to better collaborate and bond with each other as members.

What's next for rehab.ai

1) Acquire consent from other institutions, which will be used to train our model 2) Make improvements to our program until statistically significant and impactful 3) Finish building the mobile app

Built With

• css
• flask
• html
• javascript
• matplotlib
• numpy
• pandas
• python
• tensorflow