Facial Recognition System

A Python-based facial recognition system using face_recognition and OpenCV. The system allows you to:

1. Capture headshots for each person.
2. Train a model to recognize faces.
3. Run a live facial recognition session using your webcam.

---

Folder Structure

facial-recognition/ ├── dataset/ # Folder to store headshots of each person │ ├── Name1/ │ ├── Name2/ │ └── ... ├── headshots.py # Script to capture photos of individuals ├── train_model.py # Script to train the facial recognition model ├── facial_recognition.py # Script to run live recognition ├── encodings.pickle # Generated model file (after training) ├── requirements.txt # Dependencies └── README.txt

Important: Create a folder called dataset before starting. Inside it, create a folder for each person you want to recognize (e.g., dataset/Mike).

---

Setup

1. Clone the repository: git clone https://github.com/Manolis8/facial-recognition.git cd facial-recognition

2. Install dependencies: pip install -r requirements.txt

---

Step 1: Capture Headshots

Run headshots.py to take pictures of each person:
python headshots.py

• Replace the variable name with the person’s name.
• Press SPACE to capture a photo.
• Press ESC to exit.
• Tip: Take 100+ photos per person from multiple angles, lighting conditions, and expressions.

All images are saved in dataset//.

---

Step 2: Train the Model

Once you have enough images, run train_model.py to generate facial encodings:
python train_model.py

This will:

• Read all images in the dataset folder.
• Detect faces and generate embeddings.
• Save the encodings to encodings.pickle.

---

Step 3: Run Live Facial Recognition

Run facial_recognition.py to start detecting faces in real-time: python facial_recognition.py

• The system uses the webcam and identifies people from encodings.pickle.
• Recognized faces will display their names on screen.
• Press Q to quit.

---

Tips for Better Accuracy

• Take many photos from different angles, expressions, and lighting conditions.
• Avoid blurry images.
• Keep the background simple for training.
• Make sure each person has their own folder in dataset/.
• Re-train the model after adding more images.

---

How it Works

1. headshots.py – Captures images of individuals.
2. train_model.py – Processes images, extracts face encodings, and serializes them to encodings.pickle.
3. facial_recognition.py – Uses webcam feed to detect and recognize faces in real-time using the trained encodings.

---

Notes

• Make sure dataset/ exists and has at least a few images per person before training.
• The larger and more varied your dataset, the better the recognition accuracy.
• Designed for educational and demo purposes.