Traffic Vision: AI-Powered Traffic Monitoring System and Signal Optimization

Urban traffic is unpredictable. Congestion spikes during peak hours, emergency vehicles get stuck, and intersections often operate on outdated static timers.

So, we asked ourselves:

Can AI help manage traffic more intelligently, in real time?

That’s what led to the creation of Traffic-Vision — an AI-powered traffic monitoring and signal optimization system. Using YOLOv8, real-time video analysis, and adaptive traffic control, we built a platform that helps optimize urban mobility, improve road safety, and offer real-time insights.

🧠 The Idea

Most traffic lights today still operate on fixed cycles, regardless of how congested or empty a junction is. That inefficiency inspired a core question:

What if traffic signals could adapt based on live road conditions?

Traffic-Vision was built to answer that — by detecting live congestion, recognizing emergencies, and adjusting traffic flow dynamically.

🔍 Key Features at a Glance

• 🚘 Real-time vehicle & pedestrian detection


• 🚑 Emergency vehicle and accident detection (powered by YOLOv8)


• 🔁 Adaptive signal control based on traffic load


• 🔥 Heatmap-based congestion visualization


• 📩 Telegram alerts for critical events


• 📊 Dashboards for live and historical traffic metrics

🧩 System Architecture: How It Works

At its core, Traffic-Vision processes video streams through a smart computer vision pipeline and wraps that intelligence in a clean, modular interface.

Here’s the typical flow:

1. 
   Live video feeds are ingested from traffic cameras or footage.


2. 
   YOLOv8 models detect objects like cars, bikes, Bicycle, Truck, pedestrians, ambulances, firetruck and accidents.


3. Based on the real-time analysis, the system:
   
   
   
   
   • Adjusts virtual traffic signal logic
     
   
   
   • Generates heatmaps
     
   
   
   • Sends automated alerts
     
   
   
   • Logs all analytics into an SQLite database
   
   
   
   


4. A Streamlit dashboard visualizes everything — from congestion zones to emergency events.

🧪 Under the Hood: Tech Stack Breakdown

🤖 Machine Learning

• 
  YOLOv8m: For object, emergency vehicle, and accident detection


• Multiple detection models tailored for high accuracy & moderate speed

🧱 Backend

• 
  PyQt6: Graphical user interface for managing zones and controls


• 
  SQLite: Light, file-based DB for storing traffic data


• 
  Streamlit: Interactive dashboard for data analytics

💬 Messaging

• 
  Telegram Bot API: Sends real-time alerts on crashes or emergencies

🛠️ Step-by-Step Functionality

1. 🎯 Monitoring Zones

• Users draw zones over any video feed: road lanes, sidewalks, intersections.


• These zones help classify traffic density and enable fine-grained analytics.

2. 📦 Inference in Action

• 
  

  Real-time detection overlays show:

  
  
  
  
  
  • Vehicle/pedestrian counts
  
  
  • Congestion levels
  
  
  • Emergency vehicle detections
  
  
  • Accident alerts
  
  
  
  
  
  

3. 🚦 Adaptive Signal Control

• 
  

  Virtual signals respond to live data:

  
  
  
  
  
  • Congested zones get longer greens
  
  
  • Routes are cleared for ambulances or fire trucks
  
  
  • Accidents trigger instant alerts
  
  
  
  
  
  

4. 📊 Data & Dashboards

• All events are stored with timestamps.


• 
  

  Streamlit dashboards visualize:

  
  
  
  
  
  • Heatmaps
  
  
  • Zone-specific metrics
  
  
  • Incident logs
  
  
  
  
  
  

💻 Hardware & Performance Notes

Though no physical hardware is required, this system processes video at real-time or near real-time speeds, so we recommend:

• 
  GPU acceleration: NVIDIA (CUDA ≥ 12.4) or Apple Silicon (M2)


• 
  RAM: Minimum 8 GB, ideally 16 GB+


• 
  Python: Version 3.9 or newer

🔬 What We Learned

Here are some insights from development:

✅ Spatial context matters: Smart zoning leads to smarter decisions.

✅ Model size vs. speed: YOLOv8m hit the sweet spot in terms of balance.

✅ Visualization is key: Congestion data must be understandable at a glance.

✅ Real-time feedback loops elevate traffic systems from passive to adaptive.

📦 Try It Yourself

Get started with just a few commands:

git clone https://github.com/Wydoinn/Traffic-Vision

cd Traffic-Vision

pip install -r requirements.txt

python app.py

Then launch the dashboard with:

streamlit run visualizer.py

📎 GitHub: github.com/Wydoinn/Traffic-Vision

💬 Feedback or ideas? Drop a comment — I’d love to connect!