Sit Up!

Inspiration

While we were coming up with ideas on what to make, we looked around at each other while sitting in the room and realized that our postures weren't that great. You We knew that it was pretty unhealthy for us to be seated like this for prolonged periods. This inspired us to create a program that could help remind us when our posture is terrible and needs to be adjusted.

What it does

Our program uses computer vision to analyze our position in front of the camera. Sit Up! takes your position at a specific frame and measures different distances and angles between critical points such as your shoulders, nose, and ears. From there, the program throws all these measurements together into mathematical equations. The program compares the results to a database of thousands of positions to see if yours is good.

How we built it

We built it using Flask, Javascript, Tensorflow, Sklearn.

Challenges we ran into

The biggest challenge we faced, is how inefficient and slow it is for us to actually do this. Initially our plan was to use Django for an API that gives us the necessary information but it was slower than anything we’ve seen before, that is when we came up with client side rendering. Doing everything in flask, made this project 10x faster and much more efficient.

Accomplishments that we're proud of

Implementing client side rendering for an ML model Getting out of our comfort zone by using flask Having nearly perfect accuracy with our model Being able to pivot our tech stack and be so versatile

What we learned

We learned a lot about flask We learned a lot about the basis of ANN We learned more on how to implement computer vision for a use case

What's next for Sit Up!

Implement a phone app Calculate the accuracy of our model Enlarge our data set Support higher frame rates

Built With

• flask
• javascript
• sklearn
• tensorflow