CivicLens

Home screen
scanning hazard
Google gemini detected hazard and give it a high severity and create a ticket
deduplication engine: if another user upload same hazard the ticket will be merged with previous one and upvote it giving it high priority
all reports are stored here
if we click on verify fix it will open live camera again and gemini will verify fix
fix verified
if we click on notify it will draft an email automatically to send to higher authorities

Inspiration

The inspiration for CivicLens stems from the persistent gap between civic complaints and municipal action. In many cities, reporting a pothole feels like shouting into a void. We wanted to build a system where AI-driven accountability ensures that issues are accurately categorized and systematically tracked until they are resolved.

What it does

CivicLens is an AI-powered civic platform designed to streamline the lifecycle of municipal issues:

Intelligent Reporting: Citizens upload photos, and Gemini 2.5 flash identifies the issue type and severity, ensuring the right data reaches the right department.

Visual Deduplication: Using Gemini’s multimodal reasoning, the app detects if a problem has already been reported nearby, preventing database clutter.

Escalation Engine: The system includes a built-in mechanism to escalate unresolved issues. It uses Gemini to draft formal, evidence-backed reports that can be sent to higher authorities if deadlines are missed.

Verified Proof of Work: To ensure transparency, the app's logic requires "Live Scanning" at the exact GPS coordinates to verify that a fix has actually been implemented.

How we built it

Frontend: Developed using Flutter for a smooth, cross-platform mobile experience.

AI Engine: Integrated the Gemini API to handle image reasoning, data structuring, and professional correspondence drafting.

Backend: Firebase for real-time database management and Google Maps API for location-locked verification.

Challenges we ran into

A major technical challenge was creating a robust Deduplication system. We had to train the logic to distinguish between two different potholes and the same pothole from different angles. We addressed this by combining Gemini’s visual landmark analysis with GPS proximity thresholds.

Accomplishments that we're proud of

We are proud of building a functional Escalation Engine that can take a stale complaint and instantly transform it into a professional, structured document ready for municipal oversight.

What we learned

We learned how to implement multimodal AI workflows, specifically how to use Gemini to bridge the gap between unstructured real-world images and structured database entries.