Face Detection & Recognition 🎭

Real-time face detection, facial feature recognition, and face identification in images, videos, and live webcam feeds. Perfect for security systems, attendance tracking, or your next AI project!

---

What It Does 🎯

• Face Detection 👤: Find and highlight all faces in images, videos, or webcam streams
• Facial Features 👁️: Detect eyes, nose, mouth, and other facial landmarks
• Face Recognition 🔍: Identify people by comparing against known faces
• Multiple Formats 📁: Works with images, videos, and live webcam feeds
• Easy API 🔧: Simple Python class you can use in your own projects

---

Quick Start (5 mins) 🚀

Step 1: Install Dependencies

pip install -r requirements.txt

What gets installed:

• opencv-python — Computer vision library
• numpy — Numerical computing
• matplotlib — Visualization
• scikit-learn — Machine learning

Step 2: Run Face Detection

python Face\ Detection/facedetect.py

Step 3: Use in Your Code

from FaceDetect import FaceDetect

# Initialize with image
fd = FaceDetect(source='image', image_path='photo.jpg')

# Detect faces
faces = fd.detect_faces()
print(f"Found {len(faces)} faces!")

# Recognize faces (requires known_faces folder)
identified = fd.recognize_faces(known_faces_folder='./known_people/')
for person, location in identified:
    print(f"Found {person} at {location}")

---

Features ✨

Face Detection 🎭

✅ Detect all faces in images
✅ Real-time webcam processing
✅ Video file processing
✅ Mark faces with rectangles
✅ Extract individual face crops
✅ Confidence scores

Face Recognition 🔎

✅ Identify known people
✅ Name labels on detected faces
✅ Confidence matching
✅ Handle multiple angles
✅ Batch processing

Facial Features 👁️

✅ Eye detection
✅ Nose detection
✅ Mouth detection
✅ Face landmarks
✅ Feature drawing

---

How to Use 📖

1. Detect Faces in an Image

python Face\ Detection/facedetect.py --image photo.jpg --output detected.jpg

Output: New image with faces highlighted

2. Process Webcam (Live)

python Face\ Detection/facedetect.py --webcam --show-features

Features: Real-time detection with facial landmarks

3. Recognize Faces

python Face\ Detection/facedetect.py --webcam --recognize --known-faces ./people/

Features: Identifies people in real-time

4. Process Video File

python Face\ Detection/facedetect.py --video my_video.mp4 --recognize

Output: New video with faces labeled

---

Project Structure 📁

Face-Detection/
├── Face\ Detection/
│   └── facedetect.py           # Main FaceDetect class
├── cascades/                   # Pre-trained detection models
├── known_faces/                # Reference images for recognition
├── resources/                  # Sample images
├── output/                     # Generated output files
└── README.md

---

Code Examples 💻

Example 1: Simple Face Detection

import cv2
from FaceDetect import FaceDetect

# Load image
image = cv2.imread('photo.jpg')

# Initialize detector
fd = FaceDetect()

# Detect faces
faces = fd.detect_faces(image)

# Draw rectangles
for (x, y, w, h) in faces:
    cv2.rectangle(image, (x, y), (x+w, y+h), (0, 255, 0), 2)

# Save result
cv2.imwrite('detected_faces.jpg', image)
print(f"Found {len(faces)} faces!")

Example 2: Real-time Face Recognition

from FaceDetect import FaceDetect
import cv2

# Initialize
fd = FaceDetect(source='webcam')

# Start webcam
cap = cv2.VideoCapture(0)

while True:
    ret, frame = cap.read()
    
    # Recognize faces
    identified = fd.recognize_faces(frame, 'known_faces/')
    
    # Draw labels
    for name, (x, y, w, h) in identified:
        cv2.rectangle(frame, (x, y), (x+w, y+h), (0, 255, 0), 2)
        cv2.putText(frame, name, (x, y-10), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (255, 0, 0), 2)
    
    cv2.imshow('Face Recognition', frame)
    
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break

cap.release()
cv2.destroyAllWindows()

---

Requirements 📦

opencv-python>=4.5.0
numpy>=1.19.0
matplotlib>=3.3.0
pillow>=8.0.0
scikit-learn>=0.24.0
scipy>=1.5.0

Install all:

pip install -r requirements.txt

---

Tips & Tricks 💡

Better Recognition

• Use multiple clear photos of each person (different angles, lighting)
• Good lighting improves detection significantly
• Photos should be face-focused (close-ups work best)
• Use at least 3-5 photos per person

Faster Processing

• Lower image resolution for real-time video
• Use GPU acceleration if available
• Process every 2nd frame instead of every frame
• Crop regions of interest before processing

Better Accuracy

• Use high-quality images
• Ensure good lighting conditions
• Clean up known_faces folder (remove blurry/bad photos)
• Test with multiple threshold values

---

Common Use Cases 🎬

✅ Security Systems — Door entry, surveillance
✅ Attendance Tracking — Automatic classroom/office attendance
✅ Photo Organization — Auto-tag people in photos
✅ Video Analysis — Count people, detect crowds
✅ Access Control — Secure building/room entry
✅ Social Media — Auto-face blur for privacy
✅ Retail Analytics — Customer counting & analysis

---

Troubleshooting 🔧

Q: "No faces detected in my image"

• Check image quality and lighting
• Try a closer/clearer photo
• Adjust detection threshold in code
• Ensure face is clearly visible

Q: "Recognition not working"

• Add more training images (minimum 3-5 per person)
• Use clear, well-lit photos
• Remove low-quality images from known_faces/
• Increase similarity threshold

Q: "Slow performance"

• Reduce video resolution
• Process fewer frames per second
• Use GPU acceleration (CUDA)
• Process only face regions

Q: "Wrong faces detected"

• Check cascade file is correct
• Adjust scaleFactor and minNeighbors
• Try different detection method
• Use higher resolution image

---

Advanced Features 🚀

Custom Cascade Classifiers

Train your own detection model:

opencv_traincascade -data cascade_data/ -vec samples.vec -bg negatives.txt -numStages 15

GPU Acceleration

Enable CUDA for faster processing:

cv2.cuda.setDevice(0)  # Select GPU device

Multi-threaded Processing

Process multiple streams simultaneously for better performance

---

What's Next? 🔮

Features to Add

• Deep learning-based detection (YOLO, SSD)
• Age/gender estimation
• Emotion detection
• Face beautification filters
• Real-time video streaming
• Web interface
• Mobile app

To Deploy

• Deploy as REST API
• Create web dashboard
• Build desktop application
• Mobile app version
• Cloud deployment

---

License 📜

MIT License - Use freely, modify, share!

---

Made By

Rushitha
Computer Vision | AI | Open Source

Have feedback? Open an issue
Want to contribute? Submit a PR

Last Updated: February 2026
Status: ✅ Active & Maintained