CNN Based Smart Traffic Signal Control Using Lane-Wise Vehicle Detection Using YOLOv8

Overview

This project uses YOLOv8, a CNN-based object detection model, to detect vehicles from traffic images and dynamically allocate green signal timings based on lane-wise vehicle density.

Features

• Vehicle Detection using YOLOv8
• Lane-wise Vehicle Counting
• Traffic Density Estimation
• Dynamic Signal Timing

Dataset

Traffic Vehicle Object Detection Dataset (Kaggle)

Methodology

1. Train YOLOv8 on traffic images.
2. Detect vehicles from road images.
3. Divide image into three lanes.
4. Count vehicles in each lane.
5. Assign green signal duration.

Formula:

Green Time = 15 + 2 × Number of Vehicles

Results

• Precision: 0.847
• Recall: 0.953
• mAP50: 0.942
• mAP50-95: 0.617

Technologies Used

• Python
• YOLOv8
• OpenCV
• PyTorch
• Ultralytics

Author

Swaroop Mudholkar