Oculon

Inspiration

Our inspiration for Oculon came from a desire to make emergency services more efficient. Every year, countless lives are lost due to accidents that are not reported in time. Our team set out to change this reality by utilizing real-time computer vision and AI powered image detection. Oculon can instantly detect, verify, and report traffic accidents as they occur — ensuring help is on the way when every second counts.

What it does

Oculon is an AI-powered traffic incident detection system that scans road camera footage and automatically identifies and verifies accidents and emergencies. The instant a verified crash occurs, Oculon gives you the option to send an emergency notification (call) to a first responder.

How we built it

Oculon was built as a full-stack AI system combining real-time computer vision, cloud automation, and a modern web interface. The frontend was developed using React.js and Tailwind CSS, featuring a responsive dashboard that streams live video feeds, displays detection overlays, and logs alerts in real time. The backend was built with Python (FastAPI) to handle video processing, AI inference, and API communication. For detection, OpenCV used to track moving vehicles and detect sudden collisions, while Google Gemini’s Vision-Language Model (VLM) provided deeper semantic understanding — verifying whether a detected event was an actual accident. Once verified, the backend triggered AWS SNS to automatically send alerts to nearby hospitals and police stations. Additional AWS services like S3 were used for storing frames and alert evidence, and DynamoDB maintained alert logs and system metadata.

Challenges we ran into

Optimizing the AI model to accurately detect and verify an accident when it happens instead of throwing false positives.
Properly implementing a calling functionality through Twilio. ## Accomplishments that we're proud of
Creating a modern, intuitive, and readable user interface.
Developing a way for an AI model to detect a valid crash at the right time and integrating that within our user interface. ## What we learned
We learned how to integrate both backend and frontend to create a functioning full-stack application.
Learned the importance of having a modern and readable user interface. ## What's next for Oculon
Additional scalability by adding functionality for accidents apart from just car crashes.
Higher quality security and encryption.