🎯 Face Tracking with OpenCV

Gerçek zamanlı yüz takip sistemi - OpenCV kullanarak webcam'den canlı yüz takibi yapın. CSRT, KCF ve MOSSE algoritmaları ile hızlı ve hassas takip imkanı sunar.

A high-performance, real-time face tracking system using multiple tracking algorithms (CSRT, KCF, MOSSE). Easy-to-use interface for face detection and tracking with webcam support.

Features

• 🎯 Multiple Tracker Options: CSRT, KCF, and MOSSE algorithms
• ⚡ Real-time Performance: Optimized for live webcam tracking
• 🔍 Smart Face Validation: Periodic validation to prevent drift
• 🎨 Clean API: Simple, intuitive interface
• 📹 Webcam Support: Built-in camera integration

Tracker Comparison

Tracker	Speed	Accuracy	Use Case
MOSSE	⚡⚡⚡ Fastest	⭐⭐ Good	Fast movements, low-end hardware
KCF	⚡⚡ Fast	⭐⭐⭐ Better	Recommended - Balanced performance
CSRT	⚡ Slower	⭐⭐⭐⭐ Best	Maximum accuracy, powerful hardware

Installation

# Clone the repository
git clone https://github.com/Keremdagli/FaceTracking.git
cd FaceTracking

# Install dependencies
pip install -r requirements.txt

Quick Start

Using the Module

from face_tracker import run_tracker

# Run with KCF tracker (recommended)
run_tracker(tracker_type='KCF', window_title='My Face Tracker')

Using as a Class

from face_tracker import FaceTracker
import cv2

tracker = FaceTracker(tracker_type='KCF')
cap = cv2.VideoCapture(0)

while True:
    ret, frame = cap.read()
    if not ret:
        break

    # Your tracking logic here
    # See examples for complete implementation

    cv2.imshow('Tracking', frame)
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break

cap.release()
cv2.destroyAllWindows()

Examples

The examples/ directory contains ready-to-run examples for each tracker:

# Run from project root directory
python examples/kcf_example.py    # KCF - Recommended
python examples/csrt_example.py   # CSRT - Most accurate
python examples/mosse_example.py  # MOSSE - Fastest

# Or run directly from examples directory
cd examples
python mosse_example.py

Press q to quit any example.

How It Works

1. Detection: Uses Haar Cascade for initial face detection
2. Tracking: Switches to selected tracker (CSRT/KCF/MOSSE) for real-time tracking
3. Validation: Periodically validates tracked region to prevent drift
4. Recovery: Automatically redetects face if tracking is lost

Requirements

• Python 3.7+
• OpenCV with contrib modules (opencv-contrib-python)
• Webcam

Project Structure

FaceTracking/
├── face_tracker.py              # Main module
├── haarcascade_frontalface_default.xml  # Face detection model
├── examples/
│   ├── csrt_example.py         # CSRT tracker example
│   ├── kcf_example.py          # KCF tracker example
│   └── mosse_example.py        # MOSSE tracker example
├── requirements.txt
└── README.md

API Reference

FaceTracker Class

FaceTracker(tracker_type='KCF')

Methods:

• detect_face(frame): Initial face detection
• track(frame): Track face in current frame
• validate_face(frame, box): Validate tracked region
• reset(): Reset tracker state

run_tracker Function

run_tracker(tracker_type='KCF', window_title='Face Tracking')

Parameters:

• tracker_type: 'CSRT', 'KCF', or 'MOSSE'
• window_title: Display window title

Performance Tips

• Use MOSSE for low-end hardware or fast movements
• Use KCF for general purpose tracking (recommended)
• Use CSRT when accuracy is critical
• Ensure good lighting for better detection
• Keep face within frame for consistent tracking

Troubleshooting

Camera not opening:

# Try different camera index
cap = cv2.VideoCapture(0)  # or 1, 2, 3, etc.

Tracker not found error:

pip uninstall opencv-python
pip install opencv-contrib-python

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

Acknowledgments

• OpenCV team for excellent computer vision library
• Haar Cascade classifiers for face detection