Inspiration
Studying? Work? Gaming? Entertainment? Each and every one of us spends countless hours of the day in front of our computers. Without noticing, we find ourselves sitting in a really weird position. Over time, it is easy to notice how our posture quality decreases, and we start developing long term health issues. If there was only something that could fix this issue... Until we realised that advanced trained AI models are available that track our human body keypoints! Because of our extremely large target market, we realised that our innovation has an extremely large social impact. This fueled our 24 hour long hackathon effort to develop the app!
What it does and who it helps
We believe that simplicity is key. Our app alerts the user when bad posture is detected over an adjustable time period. At the start, configure our algorithm by taking a picture of yourself with correct posture. The front camera of your setup will automatically be chosen. You can retake the picture and see how the ML model adapts to your upper body. Then, you can simply continue working on your computer as normal, while the algorithm is monitoring you in the background. All people who use their computers will benefit from our app, as it encourages good posture.
How we built it
We used Python for both Front-end and Back-end. For the UI, we decided to use Streamlit for Mock testing and its efficiency. For the Back-end, we incorporated OpenCV and MediaPipe to detect the human body keypoints through trained models. We then used coordinate vector relations compared to the initial configuration benchmark to determine whether the user is slouching or still in the correct position. We developed the app with user experience as our primary concern, and are happy with our GitHub team collaboration and conflict management.
Challenges we ran into and the resulting accomplishments that we're proud of
It took us a lot of team effort to create reliable and accurate real time predictions through mathematical models. After many suggestions we finally had a working model and are proud of our implementation. Other challenges include learning a computer vision library in a very short period of time, and effectively combining the key strengths of each member to create a result that we are all happy with. We also love that our innovation improves the life of so many people in such an easy and effective way, and are proud to be the creators behind our helpful idea.
What we learned
How crucial effective teamwork is. Git and GitHub collaboration in a hectic and fast paced work environment. Agile software development practices. OpenCV and MediaPipe libraries, including the Streamlit framework. How bad our posture is :()
What's next for Spine - AI Posture Monitoring
Recording the cumulative reminder count for each session Detailed graphs motivating user by showing progress Adding Pomodoro Timer / Focus Timer for productive work -> create and save study periods Create different user profiles (own benchmarks, progress, standard settings) Create as Chrome Extension Expand to AppStore / Google Play
Log in or sign up for Devpost to join the conversation.