Aang

A path to greener postures

Comment

Inspiration

As students experiencing the challenges of remote learning, especially extended periods studying in front of a computer, we desperately needed a solution. That's why we created Aang, a posture detection software, that will promote healthy work lifestyles. This project will benefit anyone who works remotely.

What it does

Using the webcam of our computer or laptop, the webcam will take in a video feed, calibrate, and record a comfortable and proper posture of the user. Over time, the program will compare the recorded models of the proper posture to the current posture of the user. The comparison criteria include slouching, the distance from the screen, and general body posture. It also includes a timer to recommend the user to take breaks for set intervals.

How we built it

To build the frontend, we utilized ReactJS and a React library, Chakra UI, to create a clean, polished design. For the backend, we implemented TensorFlow Posenet. Posenet takes models of processed camera images as the input and outputs information about keypoints. Keypoints are indicated through an ID, and a confidence score. Coding with JavaScript, we mathematically calculated and compared dynamically generated data sets extracted from a webcam to determine good postures from bad postures. Everything is done inside the browser to preserve user privacy, as no webcam data ever leaves the user's computer.

Challenges we ran into

Implementing TensorFlow was by far the biggest challenge. Setting up the software was troubling by itself. Although, we were able to complete the configuration and set up the starter code through patience and determination. Calculating the math was another challenge as we had to consider numerous edge cases. However, the most grueling blockade was getting the PoseNetdetection to work consistently. Accuracy was tedious to work around since there were multiple occasions of the models being read incorrectly.

Accomplishments that we're proud of

Using the TensorFlow AI was a huge milestone for us. We were able to experience AI in an approachable manner. Learning React development was another plus. This was overall a fantastic opportunity to challenge ourselves by learning Web Development and Machine Learning.

What we learned

We learned to program desirable Web Development with diverse libraries. AI was a new skill set we acquired through this project.

What's next for Aang

We would like to continue our mission to promote healthy work styles. This will be accomplished by further expanding our presence through App Development and Desktop Apps. For features, we are aspiring to add components to detect eye strains and send notifications to take breaks. Lastly, we would like to engineer a Smart Mirror integration.

Built With

Share this project:

Updates