OpenPark

Our team
Auramaxing with our elegant setup

Inspiration

Our inspiration comes from something Houstonians deal with every single day: commuting. Traffic alone is already exhausting, but parking makes it even more frustrating. After long drives, circling parking lots just to find an open spot feels like wasted time and energy. We wanted to solve a problem that we personally experience, so we started asking ourselves: what would be the best way to eliminate the stress of finding parking? That question became the foundation of our project.

What it does

Our app allows users to view parking lots from an aerial perspective and clearly see which spots are open or occupied. With this information, commuters can choose the most convenient parking spot before even entering the lot, saving time and avoiding unnecessary laps.

How we built it

We used a YOLOv11 model to detect open and occupied parking spots from aerial images. Using this detection data, we guide users directly to available spots instead of relying on luck or parking at the furthest possible location. The backend of the application was built using FastAPI, which handled requests smoothly and reliably, while the model inference powered the real-time parking availability updates.

Challenges we ran into

There were honestly more challenges than we could count. From dealing with poor-quality datasets and spending hours troubleshooting model training, to struggling with low confidence predictions and frontend glitches, the process was intense. Training a reliable model took much longer than expected, and debugging the frontend added another layer of difficulty. Surprisingly, the backend was the least of our concerns thanks to how stable and reliable FastAPI turned out to be.

Accomplishments That We're Proud Of

Although we couldn’t get the full 24-hour hackathon experience due to exams and time constraints, and only started around 6 PM, we still managed to build a working prototype in under 12 hours. Training a model, developing the backend, and creating a functional frontend within that time frame was extremely challenging but also incredibly rewarding.

What we learned

We had very limited experience building full web applications before this project, but we pushed through by learning and using FastAPI and Vultr for the backend and server respectively. This was our first time working with both tools, and we found them to be surprisingly user-friendly and powerful. More importantly, we learned how to work through uncertainty, adapt quickly, and collaborate under pressure.

What's next for OpenPark

The big idea for OpenPark is to scale it to places like airports, arenas, and universities where parking availability changes dynamically throughout the day. In these environments, a system like OpenPark could guide users directly to the best terminal, lot, or gate-specific parking based on their destination. A solution like this doesn’t currently exist at scale, and we believe it has strong real-world potential.