🧑 Face Recognition Using Deep Learning

A deep learning-based face recognition system developed using TensorFlow/Keras and the Yale Faces Dataset. The project detects faces, preprocesses facial images, trains a Convolutional Neural Network (CNN), and recognises individuals from unseen images.

---

📖 Overview

Face recognition is one of the most widely used computer vision applications, with use cases ranging from security systems to biometric authentication.

This project demonstrates an end-to-end face recognition pipeline using the Yale Faces Dataset, covering everything from data preprocessing to model evaluation and prediction.

---

🚀 Features

• Face detection using MTCNN
• Image preprocessing and normalisation
• CNN model built using TensorFlow/Keras
• Model training and validation
• Accuracy and loss visualisation
• Classification report
• Single image prediction

---

🛠 Technologies Used

• Python
• TensorFlow / Keras
• OpenCV
• MTCNN
• NumPy
• Pandas
• Matplotlib
• Scikit-learn

---

📂 Dataset

The project uses the Yale Faces Dataset.

After extracting the dataset, the folder structure should be:

yalefaces/
├── train/
└── test/

The dataset README inside the data/ folder contains additional setup instructions.

---

📁 Repository Structure

Face-Recognition/
│
├── main.ipynb
├── data/
│   ├── archive.zip
│   └── README.md

---

⚙️ Workflow

1. Load the Yale Faces dataset
2. Detect faces using MTCNN
3. Preprocess and resize images
4. Create training and validation datasets
5. Train a Convolutional Neural Network
6. Evaluate model performance
7. Predict identities from unseen images

---

📊 Results

The model is evaluated using:

• Training Accuracy
• Validation Accuracy
• Training Loss
• Validation Loss
• Classification Report
• Confusion Matrix

---

📸 Project Screenshots

Sample Face Detection

---

Training Accuracy

---

Validation Loss

---

Model Prediction

---

▶ Running the Project

Clone the repository

git clone https://github.com/NitishW27/Face-Recognition.git

Install dependencies

pip install -r requirements.txt

Open the notebook

jupyter notebook main.ipynb

Run all cells sequentially.

---

💡 Future Improvements

• Improve recognition accuracy using transfer learning
• Compare CNN performance with FaceNet and VGGFace
• Deploy as a web application using Streamlit
• Add webcam-based real-time face recognition
• Support larger face datasets

---

👨‍💻 Author

Nitish Wadpally

Master of Data Science & Decisions
UNSW Sydney

• GitHub: https://github.com/NitishW27

---

📄 License

This project is licensed under the MIT License.