vtrack - Vehicle Tracking and Counting System

A real-time vehicle detection and counting system using YOLO object detection and SORT tracking algorithm.

Overview

vtrack is a computer vision application that processes video footage to detect, track, and count vehicles in traffic. It uses state-of-the-art deep learning models to identify vehicles and sophisticated tracking algorithms to maintain unique IDs across frames, allowing accurate counting of vehicles moving in different directions.

Features

• Real-time vehicle detection using YOLO (You Only Look Once)
• Multi-object tracking with SORT (Simple, Online and Realtime Tracking)
• Direction-based vehicle counting (up/down lanes)
• Visual overlays for tracking visualization
• Video output with bounding boxes, IDs, and counts
• Configurable counting lines and detection parameters

Demo

result.mp4

Installation

1. Clone the repository:

   git clone https://github.com/BLShaw/vtrack.git
   cd vtrack

2. Install required dependencies:

   pip install -r requirements.txt

Usage

1. Place your traffic video file in the assets/ directory
2. Update the video path in main.py if needed (default is assets/traffic_cam.mp4)
3. Configure counting line positions in main.py based on your video
4. Run the application:

   python main.py

The system will:

• Detect vehicles (cars, trucks, buses, motorcycles)
• Track them across frames using unique IDs
• Count vehicles crossing defined lines
• Display real-time counts on screen
• Save the processed video as result.mp4

Configuration

Counting Lines

You can adjust the counting lines in main.py:

• line_up: Line for vehicles going in the upward direction
• line_down: Line for vehicles going in the downward direction

The format is [x1, y1, x2, y2] representing coordinates of the line.

Detection Parameters

• tracker: SORT tracker parameters (max_age, min_hits, iou_threshold)
• Vehicle classes: Currently configured for "car", "truck", "bus", "motorbike"

File Structure

vtrack/
├── main.py            # Main application logic
├── main_cli.py        # For CLI enviornments
├── sort.py            # SORT tracking algorithm
├── requirements.txt   # Python dependencies
├── README.md         # This file
└── assets/
    ├── traffic_cam.mp4  # Input video
    ├── mask.png         # Region of interest mask
    ├── graphics.png     # UI overlay graphics
    └── graphics1.png    # UI overlay graphics

Customization

• Adjust Detection Area: Modify mask.png to limit detection to specific areas
• Change Vehicle Types: Modify the vehicle class filter in main.py
• Update Counting Lines: Adjust line_up and line_down coordinates
• Change Model: Update the YOLO model in main.py (e.g., yolov8n.pt, yolov8m.pt)

Acknowledgments

• YOLO for the object detection model
• SORT algorithm for multi-object tracking
• OpenCV for computer vision operations
• cvzone for computer vision utilities