AI-based Smart Civic Complaint Classification System

About the project

Municipal offices still sort most complaints by hand: someone reads each complaint, guesses the department and forwards it. This AI civic complaint Android project removes that step. The citizen clicks a photo of the problem, the app adds the GPS location, and Google Gemini Vision decides whether it is Garbage, Road, Water, Electricity, Streetlight or Drainage, how urgent it is and how sure it is. The server then sends the complaint to the correct department and the least-busy officer of that ward, with a visible deadline.

How it works

• Citizen takes a photo (camera or gallery) and writes one line; the app reads GPS and the address
• Gemini Vision returns category, confidence %, priority, a short title and an AI summary
• If confidence is low, the app asks the citizen to confirm or change the category
• Server checks for duplicates, saves the complaint, picks the department and officer and sets an SLA
• Officer marks In Progress and can close it only with a camera proof photo taken within 200 m of the site
• Citizen sees the full timeline and proof, rates the work or reopens it within 7 days
• Without internet or API key, an offline keyword classifier keeps the app working

Users

• Citizen: reports complaints with AI, tracks them, sees nearby complaints on a map, gives +1 support, rates and reopens
• Department Officer: gets a prioritised task list with SLA time left, updates status, resolves with GPS-checked proof, corrects the AI category, downloads a performance PDF
• Admin: manages wards, departments, categories and officers, overrides AI decisions, publishes notices and prints reports

Highlights

• 3 roles, 18 Android screens, 26 PHP APIs, 14 admin pages, 12 MySQL tables
• Gemini Vision image classification with confidence, priority and offline fallback
• Automatic routing to department and least-loaded officer, SLA deadline by priority
• Proof photo with Haversine distance check, ratings, reopen, +1 support (auto High priority at 5)
• Leaflet + OpenStreetMap complaints map (no Maps API key needed)
• Validation on both app and server (email, 10-digit mobile, password, pincode, photo type and size, GPS)
• Prepared statements, password hashing, double-submit protection, friendly error messages
• Sample data with 52 complaints in 8 wards; demo logins for admin, citizen and officer

What you get

• Complete Android Studio project (Java) and PHP admin panel source code
• MySQL database file with sample data
• Project report (Word, 69 pages) with DFD, ER, UML diagrams and test cases
• 2-page project abstract (Word + PDF)
• Project explanation with 32 viva questions
• Setup guide (Word + PDF) and installation help

Who can use this project

Final year students of BE / B.Tech (Computer, IT, ENTC), MCA, BCA, M.Tech and Diploma looking for an Android project with source code and documentation. It suits students who want a Gemini AI Android project, a smart city or e-governance project, or a civic complaint app project with a real admin panel.

Frequently asked questions

• Do I need to pay for the Gemini API? No. A free key from Google AI Studio is enough for testing and demos, and the app also works offline with keyword-based classification.
• Can I run it on a real phone? Yes. Copy the PHP folder to XAMPP, import the SQL file and type your laptop's Wi-Fi IP on the app's login screen.
• Can new complaint categories be added? Yes. The admin adds a category with keywords and a department; no app code change is needed.
• Does the officer really have to be at the site? Yes. The server checks the officer's GPS against the complaint location before accepting the proof photo.