Inspiration
Daniel has functional scoliosis — bad posture turned worse because noone detected it in time. 3% of the population has scoliosis, and 80% suffer from back pain due to sedentary lifestyles. It is best treated early on before the spinal curve worsens. He turned to physical therapy, which is reactive, expensive, and infrequent. YouTube videos are on-demand but nonpersonalized. Even if a patient receives care, they have no way to track how their body is adapting and get timely feedback on the treatment. More broadly, physical healthcare is under-impacted by technology.
What it does
We want to empower everyone with healthy postures in an increasingly posture-fucked world. Using computer vision on a phone, we help people discover spinal dysfunction, suggest unique treatment plans, and monitor their progress.
How we built it
We use PoseNet, a CNN-based computer vision algorithm that can estimate pose and key body part. We build the backend using Django and SQLite, and the front end using JavaScript and Bootstrap.
Challenges we ran into
This is the first hackathon for half our team, and none of us have used Django before. It was difficult to get accustomed to the frameworks' file structures and we got stuck debugging the backend for hours.
Accomplishments that we're proud of
We are proud of making the AI work with high accuracy. We are hope that our work will empower millions of patients around the world.
What we learned
We learned how to use Django and ChatGPT effectively. Never give up debugging and use source control well.
What's next for PosturEyes
Fine tuning the PoseNet model to detect for other body parts like scapula and vertebrae. Adopting industry papers for more advanced scoliosis modeling. Introducing side measurement for cervical (neck) health monitoring . Integrating video for more accurate measurement, including prompted range of motion exercises.
Log in or sign up for Devpost to join the conversation.