EdgeAI: Vehicle Tracking & Visualization

• Berat Emir Mutlu
• Kazim Efe Koçyiğit

---

📌 Overview

This project performs real-time vehicle detection and tracking on video input using YOLOv8.
It includes:

• A vehicle tracking pipeline in Python (OpenCV + YOLOv8).
• A web-based visualization interface to display results.

The code is designed to be extendable for color detection, direction detection, and possible Federated Learning integration.

---

🧠 Current Features

Vehicle Detection & Tracking

• YOLOv8 for detecting vehicles in video frames.
• Tracking across frames using tracker.py + SORT algorithm.
• Real-time display with bounding boxes and labels.

Web Visualization

• Static HTML/CSS/JS interface to load and display prediction data from .json files.
• Data table & chart visualization (requires manually prepared JSON).

---

🚀 Quick Start

1) Requirements

• Python ≥ 3.8
• Packages:
  opencv-python, numpy, ultralytics
• Browser: Any modern browser (Chrome/Firefox/Edge)
• VS Code with Live Server extension (for visualization)

Install Python dependencies:

python -m venv .venv
# Windows: .venv\Scripts\activate# macOS/Linux: source .venv/bin/activate
pip install -U pip
pip install opencv-python numpy ultralytics

---

2) Project Structure

EdgeAI-Visual-Perception/
│
├─ Specialization/ # Core detection/tracking pipeline
│ ├─ calcDirection.py # direction detection
│ ├─ cropVehicle.py # Extracts vehicle ROIs
│ ├─ detectColor.py # color detection
│ ├─ detector.py # YOLO detection logic
│ ├─ main.py # Entry point
│ ├─ renderer.py # Renders labels/boxes on frames
│ ├─ sort.py # SORT tracking implementation
│ ├─ tracker.py # Tracker initialization/update
│ ├─ utils.py # Utility functions
│ ├─ vidoe2.webm # Sample video │ └─ yolov8n.pt # YOLOv8 weights
│
├─ docs/ # Web-based visualization
│ ├─ assets/ # Images/icons/etc.
│ ├─ data.json # Example data file
│ ├─ index.html # Main HTML page
│ ├─ script.js # Logic for loading/displaying data
│ └─ style.css # Styling for visualization
│
├─ Documentation.pdf # Additional documentation
└─ README.md

---

▶️ Run the Vehicle Tracker

Currently, the only supported arguments are:

python main.py --video path/to/video --model path/to/yolov8.pt

Arguments

• --video — Path to a video file (e.g., sample.mp4).
  Default: "vidoe2.webm" (rename if needed).
• --model — Path to YOLOv8 model weights. Default: yolov8n.pt.

Examples

# Use sample video
python main.py --video vidoe2.webm --model yolov8n.pt

---

📊 Visualize Results (Web GUI)

Steps (VS Code Live Server)

1. Open Visualisation/ in VS Code.
2. Install the Live Server extension.
3. Right-click index.html → Open with Live Server.
4. Double-click index.html if browser did not launch automatically.

---