Inspiration
In my city there are several civci-issue citizens faces like potholes on road, garbage dumped on the road, street light not working. which has to be reported to the respective authorites. while doing that there some current systems(PMC care). in which the citizen first has to upload the image of the civic issue then select the main category(out of 26), sub category(out of 164), write the description of the complaint. So what i noticed is that this process is totally manual and as an AI&DS student i can automate this whole process of Main category selection, sub category selection, writing the description of the complaint by integrating AI.
What it does
my system civic-ai.tech is a web app and has a feature "analyse image with AI" which automatically returns the Main category , sub category of the complaint and auto writes the description of the complaint based on the image uploaded by the user.
How we built it
i used HTML, CSS, JAVASCRIPT for the frontend and NODE.JS , EXPRESS.JS for the backend used the LEAFLET.JS, OPENSTREETMAP for adding the map UI in the app. created and hosted the database on the SUPABASE deployed the system on VERCEL and got the domain from get.tech for free(using student id)
Challenges we ran into
the key challenge i faced was to integrate gemini-2.5-flash model api into our system and get accurate response from the gemini-2.5-flash model in json format.
Accomplishments that we're proud of
i m proud to say that
- i built a full stack smart complaint filing web app
- successfully integrated gemini api for ai assistance built a POSTGRESQL database and deployed the app live on vercel with custom domain (civic-ai.tech)
What we learned
- traditional system which requires manual effort can be successfully automated by integrating AI.
- it requires heavy effort and dedication and planing and continuos queue of failure to solve a real world problem.
What's next for CivicAI
- multilingual support
- voice assistant
- integration with official muncipal corporation(in my city,PMC)
- ai image detection
- duplicate complaint detection
Built With
- css
- express.js
- gemini-2.5-flash
- html
- javascript
- leaflet.js
- node.js
- openstreetmap
- supabase
- vercel
- vscode
Log in or sign up for Devpost to join the conversation.