Civic Grievance Tracker - Mobile App

Overview

Android mobile app for civic issue reporting. Citizens capture photos of potholes, garbage, fallen trees—app auto-extracts GPS, compresses images, and syncs with backend platform for ML severity analysis.

Mobile-Specific Features:

• One-tap photo capture with automatic GPS embedding
• Offline-first queue with background sync
• 512KB image compression (85% reduction)
• EXIF metadata for moment-of-capture location
• Public feed with community upvoting

Backend System: See Project Clean Backend for TensorFlow CNN severity classification, priority algorithm, and municipal dashboard.

Published in International Research Journal of Engineering and Technology (IRJET)↗.

Mobile Challenges & Solutions

Challenge 1: Accurate Location Capture

Pain Point

Citizens report issues like "near the park" or "on Main Street." Field teams waste hours searching. Text addresses have typos, informal landmarks are ambiguous.

Solution

App captures GPS when photo is taken—not when submitted later. Embeds coordinates in EXIF metadata. Extracts location even if user submits hours later from home. Reverse geocoding generates street address.

Implementation: FusedLocationProviderClient requests GPS on camera open. Coordinates embedded in image EXIF. Submission extracts EXIF data, not current location. Fallback prompts manual confirmation if EXIF missing.

Impact

• 📍 100% Accuracy: For in-app camera photos
• ⏱️ Zero Search Time: Field teams navigate directly
• 🗺️ Dual Format: GPS coordinates + street address

Challenge 2: Low-Connectivity Reporting

Pain Point

Citizens discover issues in rural roads, underground parking, remote areas with poor connectivity. Real-time submission fails, issues forgotten.

Solution

Local storage as source of truth. Submissions save immediately to device with success feedback. Background service monitors network, auto-uploads when available. Exponential backoff retry, conflict resolution for duplicate offline submissions.

Implementation: SharedPreferences stores queued submissions. Background WorkManager monitors connectivity. Visual sync indicators show pending uploads. Retry logic handles transient failures.

Impact

• 📱 Zero-Connectivity: Document offline, batch-upload later
• ⚡ 96% Success: Up from 73% in 3G areas
• 🎯 34% Higher Completion: Less frustration

Challenge 3: Image Size vs Quality

Pain Point

Smartphone cameras produce 4-8MB images. Drains data plans, causes timeouts. Aggressive compression degrades quality below ML analysis threshold.

Solution

Multi-stage compression: resize to 1920x1080, JPEG quality 75. Produces 400-600KB files while preserving edges, textures for ML model. Progressive upload with chunking.

Implementation: Adaptive quality settings based on dimensions. Compression preview for user verification. Chunked multipart upload via Retrofit.

Impact

• 🗜️ 85% Size Reduction: 5MB → 512KB
• 🎯 92%+ ML Accuracy: Maintained
• ⚡ 96% Upload Success: In 3G areas

Mobile Architecture

Key Decisions: Native Kotlin for sensor access. Offline-first for field reliability. Server-side ML avoids 100MB+ model. 512KB compression for 3G networks.

Backend: See Project Clean Backend for Django REST API, TensorFlow CNN, and priority algorithm.

Key Features

🚀 One-Tap Reporting - under 30 seconds capture to submit
⚡ 512KB Compression - 85% reduction, 92%+ ML accuracy
📱 Offline Queue - Zero-connectivity reporting
📍 EXIF Geolocation - Moment-of-capture GPS
🎯 Public Feed - View, upvote community issues
🔄 Background Sync - Auto-upload when network returns

Technical Highlights

1. EXIF Location Timing

2. Image Compression Pipeline

3. Offline-First Data Flow

Tech Stack

Impact

• ⚡ 96% Upload Success in 3G (up from 73%)
• 📍 100% Location Accuracy with EXIF
• 📱 34% Higher Completion via offline-first
• 🗜️ 85% Bandwidth Reduction through compression
• 📚 Published Research: IRJET V9/i4↗

Backend Platform: See Project Clean Backend for ML severity analysis, priority algorithm, and municipal dashboard.