Poshchure

hardware!

Inspiration

As computer science students, we often find ourselves glued to our screens for long hours, leading to poor posture and consequent physical discomfort or pain. Witnessing the prevalence of this issue among our peers inspired us to create Poshchure. We envisioned a solution that leverages technology to promote healthier habits and improve the well-being of individuals who spend significant time at their desks.

What it does

Poshchure is an innovative application designed to monitor and correct your posture in real-time. Using advanced machine learning algorithms and computer vision techniques, Poshchure analyzes your sitting posture through your webcam and provides instant feedback. When it detects that you’re slouching or sitting incorrectly, it sends "gentle" reminders via a wearable device to adjust your posture, helping to prevent the onset of posture-related pain and injuries.

How we built it

Poshchure was built using a combination of modern technologies. We utilized MediaPipe for real-time pose detection, which allowed us to capture and analyze posture through the webcam. Our machine learning models were developed with Scikit-learn to accurately identify and classify different posture states. The backend was developed in Python using Flask, which handles the communication between the frontend and the machine learning models. For the frontend, we used React to create a user-friendly interface that provides seamless interaction and immediate feedback.

We utilized a ESP8266 microcontroller for the wearable device. The firmware was written in C++ and it communicates with the backend through a custom UDP protocol.

Challenges we ran into

Throughout the development process, we encountered several challenges. Integrating real-time pose detection with our machine learning models required careful calibration to ensure accuracy and responsiveness. We also faced difficulties in creating a user interface that was both informative and non-intrusive. Additionally, ensuring the application ran smoothly across different devices and environments was a significant hurdle that required extensive testing and optimization.

Also, centering divs in react is impossible. And more teammates would've been great...

Accomplishments that we're proud of

Despite the challenges, we are proud of several key accomplishments. Successfully integrating MediaPipe with our custom machine learning models to achieve real-time posture detection was a major milestone. We also take pride in the intuitive and clean user interface we developed, which enhances the user experience. Moreover, creating an application that addresses a common health issue and has the potential to improve the quality of life for many individuals is something we are particularly proud of.

We also thought the integration of hardware in our project was pretty cool! It made everything more engaging.

What we learned

Through this project, we gained valuable insights into several areas. We deepened our understanding of machine learning and its practical applications in computer vision. We also learned how to effectively integrate various technologies to create a cohesive and functional product. Additionally, the importance of user-centered design and continuous testing became evident as we worked to refine the application.

What's next for Poshchure

Looking ahead, we plan to expand Poshchure’s capabilities. Future enhancements include incorporating personalized posture improvement plans and integrating with wearable devices for more accurate posture monitoring. We also aim to add features like detailed posture analytics and progress tracking to provide users with deeper insights into their habits. Finally, we envision Poshchure becoming a comprehensive wellness tool, extending its functionality to include exercises and stretches tailored to individual needs.