Pigeon

Pigeon

Inspiration 

Parking on campus can be an absolute nightmare. Too often, we waste valuable time circling crowded lots, praying a spot will open up. That frustration sparked the idea behind Pigeon. Our mission? Streamline campus parking for students, faculty, and visitors by using cutting-edge AI to provide real-time updates and seamless payment options—all in one platform.

What It Does 

Pigeon harnesses the power of real-time camera feeds to detect parking availability across campus. Our AI algorithm pinpoints which spots are open and recommends the quickest route, complete with navigational guidance. Even better, you can pay for parking right from our web app, eliminating the need to fumble with meters or kiosks. On top of that, Pigeon integrates OCR and license plate detection to help campus security identify stolen vehicles—giving you peace of mind while ensuring a safer parking experience.

How We Built It

Machine Learning Pipeline: Developed with Flask and Python for our backend, leveraging PyTorch and OpenCV for robust image detection and analysis.
Dataset & Model: Trained on datasets from Roboflow via YOLOv8, fine-tuned specifically for real-time parking space identification and license plate recognition.
Frontend & Integration: Built with HTML, CSS, and JavaScript, communicating with the backend through Google Maps API for dynamic routing and satellite mapping features.

Challenges We Ran Into

Lengthy Training Time: Fine-tuning our model took over three hours, pushing our timeline and testing our patience.
Model Integration: Combining the OCR and license plate detection features proved trickier than expected, and we ran short on time to merge them seamlessly.
A literal power outage mid-development was pretty fun haha.

Accomplishments That We’re Proud Of

High-Accuracy Models: We successfully trained and tested two separate ML models, each with around 94% accuracy.
Streamlined UI: Our user-friendly frontend, integrated with Google Maps, delivers real-time parking data and easy navigation for an elevated user experience.
Robust Data Foundation: Discovering Roboflow and relevant datasets lets us build a reliable solution that can be scaled and improved upon.

What We Learned

Importance of Ideation: Laying out a clear vision and goals upfront significantly streamlined our decision-making process.
Strategic Tech-Stack Planning: Properly scoping out our tools and frameworks ensured smoother integration down the line.
Deep Dive into APIs: Mastering Google Maps API—especially its satellite imaging and mapping capabilities—opened new possibilities for future expansion.

What’s Next for Pigeon 

Our ambition doesn’t stop at campus borders. We see a future where Pigeon revolutionizes urban parking, helping cities worldwide reduce traffic congestion and improve driver satisfaction. By scaling our technology, collaborating with municipalities, and refining our AI models, Pigeon aims to become the go-to platform for parking availability, security, and payment—transforming the way people navigate urban landscapes.