Right Angle is a full-stack application that I and a team of 3 others created during SacHacks IV. Right Angle is a posture correction web application that uses computer vision to detect whether or not a users has good or bad posture and is perfect for remote workers and students! Right Angle is a program that corrects a user's posture by identifying their body's pose and informs the user whether or not their posture should be corrected. The program first tells the user to sit up straight while it calibrates. Once calibrated, the program will signal to the user if their posture is good or bad!

In the age of COVID-19, we realized the importance of office ergonomics as we transitioned to remote work and learning. As current co-op interns, it's easy for us to forget how to maintain a healthy posture, as we spend the majority of our days seated at our desk. That's why we were inspired to create Right Angle; an application to help office workers and students create the healthy habit of correcting their posture!

Our project uses a Flask backend, and OpenCV to monitor the user's position and posture. Our frontend was made using React and Tailwind CSS.

As far as the actual slouch detection, this was accomplished using the OpenCV library. Using the library and media pipeline's facial/body detection, we use the calibrated posture to calculate the whether the user starts to slouch. This is done by cross referencing data about the relative position of the user's face, the angle of the user's neck, as well as their proximity to the camera. Multiple datapoints are required, as we cannot enforce where the user has positioned their webcam in order to check. The program continually takes an average of the user's posture as the program runs, to avoid small lapses into bad posture being reported.

Challenges

The biggest challenge when creating right angle was connecting our frontend and our backend. This was due to a number of reasons, but what it boils down to was that it was a 24 hour event, and so the front and backends were written by different people who did not have good undersatnding of how the other's worked. Additionally, our team developed on different devices with different operating sytems and architecture.

Future Improvements

In the future, and expansion of right angle would be to fully implement a backend database with accompanying data visualization, to allow users to track improvements in their sitting posture. This was intended to be one of the features of right angle, however as this was a 24 hour hackathon we were unable to implement it.