🚦 CivicPulse Bengaluru

Next-Generation AI Pothole Intelligence & Municipal Accountability Ledger

See the problem. Identify responsible contractors. Audit repairs with forensic computer vision.

---

📌 Executive Overview

Bengaluru's recurring road surface crisis is not merely an asphalt durability challenge—it is an algorithmic prioritization and contractual accountability failure:

• ⚡ Subjective Repair Dispatch: Legacy portals prioritize potholes based on VIP movement, social media virality, or political pressure rather than physical vehicle hazard metrics.
• 📦 Duplicate Ticket Spam: A single crater on the Outer Ring Road generates dozens of separate, unlinked grievances—clogging ward engineering queues and diluting emergency response.
• 📜 Evading Contractor Liability (Clause 45.2): Standard Karnataka PWD and BBMP road contracts enforce a mandatory 36-Month Defect Liability Period (DLP). When asphalt fails prematurely, private contractors routinely evade rectification notices, burdening taxpayers with secondary patchwork bills.
• 🌧️ Unverified Temporary Patching: Potholes are cosmetically filled with loose cold-mix gravel that washes away in monsoon rains, with municipal tickets prematurely marked "RESOLVED" without forensic proof.

CivicPulse Bengaluru solves this end-to-end through automated computer vision telemetry, multimodal voice-to-text NLP triage, geospatial parcel contract mapping, and automated before/after repair verification.

---

⚡ 7-Stage Intelligence Pipeline

graph TD
    A[Citizen Media: Photo / Dashcam / Voice / Text] --> B[01. Neural Detection & Boundary Segmentation]
    B --> C[02. Physical Dimension Metric Engine: Depth cm, Area m², Volume L]
    C --> D[03. 2-Stage Deduplication: Spatial Proximity & Visual Embeddings]
    D --> E[04. Geospatial Responsibility Mapping: PWD Tender & Clause 45.2 DLP]
    E --> F[05. Dynamic Hazard Priority Function: Traffic, Corridor Speed, Schools]
    F --> G[06. Multi-Agency SLA Ticket Dispatch: BBMP / BMRCL / BWSSB]
    G --> H[07. Forensic Before/After Repair Verification Audit]

Loading

---

🎨 Neo-Brutalist Design System

CivicPulse is styled with a distinct, high-density Neo-Brutalist Command Center interface engineered for municipal operations:

Token	Hex Code	Visual Swatch	Operational Function
Sage Canvas	#CFE8D6	🟢	Primary soft tint, header surfaces, pill badges & elevation highlights
Obsidian Ink	#121210	⚫	High-contrast borders (3px solid), brutalist shadows & typography
Forest Verifier	#2E8C42	🟩	Confirmed detections, verified repairs & DLP warranty compliance
Amber Warning	#E8A030	🟧	High hazard severity, NLP speech processing & SLA countdowns
Crimson Alert	#C03A3A	🟥	Critical emergency craters, active recording & SLA breaches

---

🌟 Core Modules & Capabilities

🎙️ 1. Multimodal Voice & Text Intake Cockpit

• Real-time Neural STT: Captures spoken Kannada/English complaints with continuous speech recognition (en-IN), preserving verbatim spoken context without hardcoded overwrites.
• Corridor NLP Resolution: Automatically identifies arterial corridors (Outer Ring Road, Indiranagar 100ft Rd, Whitefield ITPL, Koramangala 80ft Rd, MG Road, Jayanagar, HSR Layout, etc.) and routes to responsible civic authorities:
  • 🏛️ BBMP: Major Roads Division & Ward Infrastructure
  • 🚇 BMRCL: Metro construction covenants & Phase 2A alignments
  • 💧 BWSSB: Pipeline excavations & drainage restoration cuts
  • ⚡ BESCOM: Underground HT cable trenches & utility reinstatement

🛰️ 2. GIS Radar

• Real-time spatial radar displaying all reported craters mapped across 198 BBMP Ward boundaries.
• Layer toggles for arterial road corridors, contractor warranty zones, duplicate clusters, and priority heatmaps.
• Monocular stereoscopic telemetry rendering coordinates, elevation, and chainage.

🧠 3. Edge Computer Vision Lab

• Real-time crater bounding polygon localization with confidence scoring.
• Quantitative damage estimation calculating subbase void area ($m^2$), cavity depth ($cm$), and aggregate volume ($L$) compliant with IRC-SP-100 standards.
• Risk evaluation modeling impact across two-wheelers, auto-rickshaws, and public transit corridors.

🔍 4. 2-Stage Spatial & Visual Deduplication

• Stage 1 (Geospatial Gating): Groups candidate reports within a 25-meter radial radius.
• Stage 2 (Visual Cosine Similarity): Compares feature embeddings across camera angles ($>85%$ threshold) to merge repetitive grievances into a single consolidated Master Dossier.

⚖️ 5. Contractor DLP Ledger & Legal Notice Automation

• Queries public municipal contract packages and tender IDs associated with the defect coordinates.
• Automatically verifies whether the road is under active Defect Liability Period (Clause 45.2).
• Generates legal Defect Rectification Notices addressed to contractor directors with automated SLA countdowns.

🔬 6. AI Before/After Repair Verification

• Dual-image photographic audit comparing the original crater against contractor repair photos.
• Computes surface area reduction percentage and calculates an IRC-SP-100 Surface Smoothness Index (0–100) to prevent substandard gravel filling.

🌌 7. Interactive 3D Scroll-World Flight

• High-performance spatial 3D flight camera navigating through Bengaluru's highway network.
• Interactive 3D camera waypoints illustrating edge AI inspection along the city's tech corridors.

---

📊 Algorithmic Priority Formulation

CivicPulse eliminates subjective political bias through an objective multi-factor hazard function:

$$\text{Priority Score} = w_1 \cdot S_{\text{severity}} + w_2 \cdot T_{\text{traffic}} + w_3 \cdot V_{\text{vulnerable}} + w_4 \cdot C_{\text{speed}} + w_5 \cdot U_{\text{citizen}}$$

• $S_{\text{severity}}$ (35%): Depth in cm, crater area, and loose aggregate risk
• $T_{\text{traffic}}$ (25%): Arterial vehicle density (ORR, Hosur Road, Airport Expressway)
• $V_{\text{vulnerable}}$ (20%): Proximity to schools, hospitals, and metro interchange nodes ($<200\text{m}$)
• $C_{\text{speed}}$ (10%): Corridor design speed limit ($>60\text{ km/h}$)
• $U_{\text{citizen}}$ (10%): Verified citizen upvotes & emergency crowd confirmations

---

🛠️ Technology Stack

Frontend Architecture
├── Framework:           React 19 (Hooks, Concurrent Mode)
├── Language:            TypeScript 5.7 (Strict Null Safety)
├── Build System:        Vite 8.3 (Hot Module Replacement, Rolldown Optimization)
├── CSS & Styling:       Vanilla CSS + Tailwind CSS (Neo-Brutalist Theme)
├── Mapping & GIS:       Leaflet + CartoDB Dark Matter Tiles
├── Authentication:      Clerk Auth (@clerk/clerk-react)
├── Icons:               Lucide React
└── Visual Effects:      Canvas Confetti, Lucide SVGs, Custom 3D Viewport

Backend & Inference Pipeline
├── API Service:         FastAPI / Python 3.11 (Uvicorn)
├── Vision Processing:   OpenCV 4.x + PyTorch Contour Analysis
├── Reverse Geocoding:   Custom Bengaluru GIS Spatial Engine
└── Legal Contract DB:   BBMP TenderSURE & PWD Clause 45.2 Records

---

🚀 Quick Start Guide

Prerequisites

• Node.js v20.0+ or v22.0+
• npm v10.0+
• Python 3.10+ (optional, for local backend API)

1. Clone & Install

# Clone the repository
git clone https://github.com/skandakn/CivicPulse.git
cd CivicPulse

# Install frontend dependencies
npm install

2. Environment Configuration

Create a .env file in the project root:

# Clerk Authentication (Optional for offline demo mode)
VITE_CLERK_PUBLISHABLE_KEY=pk_test_...

# AI Speech-to-Text & Vision Keys (Optional - has resilient local fallbacks)
VITE_GEMINI_API_KEY=
VITE_GROQ_API_KEY=

3. Launch Development Server

npm run dev

Visit http://localhost:5173 (or the port shown in your terminal) to explore CivicPulse.

4. Build for Production

# Type check and build production bundle
npm run build

# Preview production build locally
npm run preview

---

🏛️ Municipal Architecture Compliance

• IRC-SP-100:2014: Indian Roads Congress code of practice for road defect inspection and automated measurement.
• Karnataka PWD Code (Clause 45.2): Standard contract conditions governing mandatory contractor Defect Liability Periods.
• BBMP Sahaya 2.0 Integration: Municipal grievance lifecycle formatting (BBMP-SHY-2026-XXXXX).

---

⚖️ License

Distributed under the MIT License. Open source for civic transparency and municipal engineering enhancement.