Transforming traffic violations into actionable intelligence.
RecLog is an AI-driven traffic intelligence platform built to help city administrators identify congestion hotspots, forecast future traffic risks, optimize resource deployment, and generate operational strategies using natural language.
Built for Flipkart Gridlock 2.0, RecLog addresses one of Bengaluru's most persistent challenges: parking-induced traffic congestion.
Bengaluru is among the world's most congested cities, with commuters losing significant hours annually due to traffic delays.
A major contributor is illegal and unregulated parking, which:
- Reduces effective road capacity
- Creates bottlenecks near junctions
- Blocks transit hubs and metro stations
- Delays emergency response vehicles
- Increases congestion ripple effects across the network
Current enforcement systems are largely reactive and depend heavily on manual monitoring.
RecLog converts raw traffic violation data into intelligent operational recommendations.
The platform enables authorities to:
✅ Identify high-risk congestion hotspots
✅ Prioritize risk using PCRI
✅ Prioritize interventions using a custom risk index
✅ Simulate future traffic conditions
✅ Optimize deployment of officers and tow trucks
✅ Generate AI-powered action plans
✅ Explain recommendations transparently
|
|
Real-time city-wide traffic monitoring with KPI cards, congestion analytics, root-cause analysis, and live incident feeds.
Interactive H3-powered hotspot visualization with PCRI risk scoring and congestion prioritization.
Natural-language decision support for resource allocation and operational planning.
|
|
|
|
Evaluate traffic interventions and forecast congestion reduction before deployment.
flowchart LR
A[Traffic Violations Data] --> B[Data Processing]
B --> C[PCRI Engine]
C --> D[Hotspot Detection]
C --> E[Dashboard Analytics]
C --> F[Simulation Engine]
F --> G[Scenario Modelling]
D --> H[Resource Allocation]
G --> I[Explainability Layer]
H --> I
I --> J[AI Traffic Copilot]
J --> K[Deployment Recommendations]
K --> L[Traffic Authorities]
At the heart of RecLog lies the Priority Congestion Risk Index (PCRI).
Unlike traditional approaches that only count violations, PCRI evaluates multiple dimensions of traffic risk.
| Component | Weight |
|---|---|
| Spatial Density | 35% |
| Violation Severity | 20% |
| Vehicle Impact | 15% |
| Repeat Offenders | 15% |
| Road Criticality | 15% |
PCRI generates a score between 0 and 100.
Higher PCRI indicates:
- Greater congestion risk
- Higher enforcement priority
- Increased resource requirements
Provides a city-wide operational overview including:
- Total Hotspots
- Critical Hotspots
- Traffic Violations
- Average PCRI
- Weekly Violation Trends
- Root Cause Analysis
- Live Incident Feed
Powered by H3 indexing and interactive GIS visualization.
Features:
- Interactive hotspot map
- Severity-based risk coloring
- Location-specific telemetry
- Critical area identification
Instantly search and filter hundreds of hotspots.
Capabilities:
- Fuzzy matching
- Real-time filtering
- Fast pagination
- Risk-based sorting
Interact with traffic intelligence using natural language.
"I have 20 officers and 3 tow trucks. Where should I deploy them on a rainy Friday?"
- Priority hotspot ranking
- Officer allocation plans
- Tow truck deployment schedules
- Predicted congestion reduction
Automatically converts PCRI scores into deployment recommendations.
| Risk Tier | Officers | Tow Trucks | Patrol Frequency |
|---|---|---|---|
| Critical | 6 | 3 | Every 15 min |
| High | 4 | 2 | Every 30 min |
| Medium | 2 | 1 | Every 1 hour |
| Low | 1 | 0 | Every 4 hours |
Evaluate interventions before deploying resources.
Supported Scenarios:
- Normal Conditions
- Rain
- Festival Traffic
- Extra Enforcement
- Traffic Diversion
Outputs:
- Predicted violations
- Projected PCRI
- Reduction efficiency
- Confidence score
Traffic authorities need trust before they act.
RecLog's Explainability Layer ensures that every recommendation generated by the platform can be understood, validated, and justified.
For every simulation, the system provides:
- 📌 Key congestion drivers
- 🌧️ Impact of environmental conditions
- 🚓 Resource allocation rationale
- 📊 PCRI component contribution
- ✅ Confidence score
- High violation density (+32%)
- Heavy vehicle concentration (+18%)
- Repeat offender activity (+11%)
- Deploy 4 Traffic Officers
- Deploy 2 Tow Trucks
- Increase patrol frequency
87%
This transforms the simulator from a black-box prediction engine into an explainable decision-support system.
A custom congestion scoring framework that evaluates traffic risk beyond raw violation counts by incorporating:
- Spatial Density
- Violation Severity
- Vehicle Impact
- Repeat Offenders
- Road Criticality
Forecasts intervention outcomes while clearly explaining:
- Why congestion levels change
- Which factors contributed most
- How recommendations were generated
- Expected confidence in predictions
Converts natural-language queries into actionable deployment strategies.
Example Query
"I have 20 officers and 3 tow trucks. Where should I deploy them on a rainy Friday evening?"
Output
- Priority hotspot ranking
- Officer allocation plan
- Tow truck deployment strategy
- Predicted congestion reduction
Uses Uber H3 Spatial Indexing to:
- Identify high-risk zones
- Visualize congestion hotspots
- Prioritize enforcement regions
- Enable location-aware decision making
Transforms congestion risk into operational recommendations through:
- Dynamic officer allocation
- Tow truck deployment planning
- Patrol frequency optimization
- Risk-based intervention strategies
- Llama 3.1
- Groq API
- Natural Language Planning
- Context-Aware Recommendations
- Scenario Modeling
- Forecasting Engine
- Risk Projection
- Intervention Analysis
- Recommendation Justification
- PCRI Breakdown Analysis
- Confidence Estimation
- Resource Allocation Reasoning
- Congestion Driver Analysis
Traffic Data
│
▼
PCRI Risk Scoring
│
▼
Simulation Engine
│
▼
Explainability Layer
│
▼
AI Traffic Copilot
│
▼
Deployment Recommendations
Most systems tell you where the problem is. RecLog tells you what to do about it.
- React
- Vite
- TailwindCSS
- React Leaflet
- Recharts
- FastAPI
- Pydantic
- Pandas
- NumPy
- Scikit-Learn
- Groq
- Llama 3.1
- OpenStreetMap
- H3 Spatial Indexing
RecLog enables:
-
Faster hotspot identification
-
Smarter resource utilization
-
Reduced traffic congestion
-
Data-driven enforcement
-
Predictive traffic planning
-
Improved incident response
-
Explainable decision-making
- Live CCTV Integration
- IoT Traffic Sensors
- Computer Vision-Based Violation Detection
- Reinforcement Learning Optimization
- Congestion Forecasting Models
- Smart City Integration
- Multi-CIty Deployment Support
git clone <repo-url>
cd RecLogcd backend
python -m venv venv
venv\Scripts\activate
pip install -r requirements.txt
uvicorn main:app --reloadcd frontend
npm install
npm run devZion
- Rishik Garg
- Disha Kaushal
- Trisha Soni
- Shresth Agarwal
Built for Flipkart Gridlock 2.0
RecLog — From Traffic Data to Smarter Decisions.