Skip to content

README.md

Latest commit

 

History

History
121 lines (81 loc) · 3.27 KB

README.md

File metadata and controls

121 lines (81 loc) · 3.27 KB
Raw

Traffic Violation Detection System

An AI-powered real-time traffic violation detection system that uses a live camera feed to automatically identify vehicle violations including helmet violations, red-light running, lane violations, overspeeding, and wrong-side driving.

Pilot selected for deployment at 3 high-traffic intersections (Q2 2026)** — with city-wide rollout potential.

What It Does

The system connects to a camera feed and uses Computer Vision to detect traffic violations in real time — no human monitoring required. When a violation is detected, it logs the event, timestamps it, and flags it for action.

Violations detected:** No helmet (two-wheeler riders) Red-light running Lane violation Overspeeding Wrong-side driving Illegal parking

Performance

Metric Result
Detection Accuracy 89%
Processing Speed 15 FPS real-time
False Positive Reduction 35% (via data augmentation)
Training Dataset 5,000+ annotated images
Labor Cost Savings ₹90L/year (manual surveillance replaced)
Intersections monitored 10+ simultaneously

Tech Stack

Component Technology
Core Model CNN (Convolutional Neural Network)
Vision Library OpenCV
Language Python
Training Framework TensorFlow / Keras
Data Processing NumPy, Pandas
Annotation Custom labeled dataset (5,000+ images)

Getting Started

Prerequisites

pip install opencv-python tensorflow keras numpy pandas

Run the detection system

python traffic_detection_single.py

Make sure your camera is connected. The system will start reading the feed and detecting violations in real time.

Project Structure

traffic-violation-detection/
│
├── traffic_detection_single.py      # Main detection script (camera input → violation output)
├── traffic_violation_detection_system.html  # Dashboard UI for monitoring & analytics
└── README.md

Dashboard

The included HTML dashboard (traffic_violation_detection_system.html) provides:

  • Live camera feed simulation across 6 intersections
  • Real-time violation analytics charts
  • Challan / fine issuing interface
  • Location-based camera monitoring
  • Hourly violation trend graphs

Open in any browser — no server needed.

Impact

This project was identified as a pilot deployment candidate for 3 high-traffic intersections, with a city-wide rollout budget of ₹8Cr. The system replaces manual surveillance, enabling:

  • 60% reduction in manual monitoring labor
  • ₹90L annual savings in surveillance costs
  • Automated, consistent enforcement across 10+ intersections

Recognition

This project was part of the work that contributed to:

  • 2nd Place — TEXPO 2K24 Innovation Pitch Competition

👤Author

Selva Ganesh K
AI Research Analyst & Technical Writer
B.Tech — Artificial Intelligence & Machine Learning

📄 License

This project is for educational and demonstration purposes.