Intelligent Traffic Management System

The project, named Intelligent Traffic Management System, controls traffic lights based on real‑time vehicle density. It automatically detects ambulances and turns the corresponding light green, supporting four separate lanes simultaneously. By predicting and counting vehicles, the system can dynamically allocate green time and includes an analysis module for traffic reduction, vehicle type distribution, and overall counts.

Project Basis and Proposed System

The foundation of the work is the paper “Intelligent Multi‑eport Vehicle Routing and Management for Smart Cities” (June 2025), which described an IoT‑based routing approach that struggled with older vehicles lacking internet connectivity. The new proposal, called Traffic Vision AI, replaces the IoT layer with deep‑learning and artificial‑intelligence techniques, focusing on object detection and autonomous traffic control.

System Functionality and Features

Traffic Vision AI calculates vehicle density for each of the four lanes and assigns traffic‑light priority accordingly—more vehicles receive a longer green phase. An ambulance detection module, trained on more than 5,000 ambulance images, instantly grants a green signal to the lane where the emergency vehicle appears. The system solves traditional problems such as static timers and manual operation that cause congestion. It requires either a live camera feed or uploaded video files and relies on the YOLO V8 algorithm for detection, together with OpenCV, Flask, and SQLite. Core features include vehicle detection and counting (cars, buses, cycles, humans, vans, trucks), ambulance detection, dynamic signal control, a real‑time dashboard, and data analytics.

Comparison with Existing Systems

Conventional traffic‑light systems use fixed timers, lack real‑time data, and depend on manual emergency‑vehicle control, resulting in low accuracy and only an 80 % traffic‑reduction rate. In contrast, Traffic Vision AI offers dynamic, density‑based control, continuous data collection, automatic ambulance priority, and high detection accuracy. Tests show a traffic‑reduction improvement from 80 % to 98 % and a reduction in average waiting periods from 30 seconds (traditional) to as low as 11 seconds with the AI system.

Proposed System Details

The architecture supports real‑time object detection via YOLO V8, enabling fully automated traffic management without human input. Ambulance detection is powered by a model trained on over 5,000 images. An analytical dashboard displays live vehicle counts, traffic‑density graphs, and alerts, while a SQLite database stores all collected information. Diagrammatic representations illustrate the flow from frame capture through vehicle and ambulance models to the user interface (login, upload, dashboard, analysis).

Project Advantages

Key benefits include dynamic signal management that adapts to current traffic conditions, immediate emergency‑vehicle priority, 24/7 autonomous operation, and detailed analysis that counts each vehicle type for precise density calculation. The system eliminates the need for manual intervention, thereby saving thousands of lives in real time.

Methodology and Implementation

The solution comprises eight models: User Authentication, Object Detection, Traffic Control, Ambulance Priority, Dashboard Analysis, Video Input/Upload, Density Calculation, and Traffic Analysis. Software requirements are Python, HTML, CSS, JavaScript, and SQLite. Hardware needs are modest—a basic i3 processor, 20 GB of disk space, and 4 GB of RAM—making the system compatible with Windows and macOS. It accepts common video formats such as MP4, AVI, and FLV.

Project Demo and Conclusion

A web‑based dashboard demonstrates the AI‑powered system in action. Users can log in, upload or stream video, and observe dynamic signal control based on lane density. The analysis page presents vehicle counts, traffic‑density graphs, and waiting‑time comparisons between traditional and AI‑driven control. A separate ambulance video showcases automatic green‑light activation with audio alerts. The source code, including YOLO V8 for vehicle detection and a specialized best.pt model for ambulances, is run via app.py. This project is positioned as an ideal final‑year engineering undertaking, offering a complete, deployable intelligent traffic management solution.