🧠 Face Recognition Using Deep Learning & Eigenfaces

A Computer Vision project by Avinash Kumar Kashyap (Roll No: 21064)
This project presents a hybrid facial recognition system combining deep learning (Siamese Network + FaceNet) and classical eigenface analysis (PCA), enhanced by DeepFace for real-world identity verification.

---

📌 Table of Contents

• 🎯 Project Objective
• 🧠 Background & Motivation
• 🗂️ Dataset Preparation
• 📊 Eigenface Analysis (PCA)
• ⚙️ Model Architecture
• 🧪 Training & Evaluation
• 🧾 DeepFace Integration
• 📈 Results & Metrics
• 📸 Visualizations
• 🚀 Future Work
• 📁 Project Structure
• 🧾 License

---

🎯 Project Objective

To build a robust facial recognition system that:

• Uses Siamese Network with FaceNet to measure similarity.
• Incorporates Eigenfaces (PCA) for dimensionality reduction.
• Employs DeepFace for robust face verification.
• Handles real-world scenarios like pose, lighting, and expressions.

---

🧠 Background & Motivation

Facial recognition is key in modern authentication systems. By combining the classical strength of eigenfaces with the learning power of neural networks, this project aims to build a highly reliable hybrid system.

---

🗂️ Dataset Preparation

📁 Data Structure

• Anchor: Base reference face
• Positive: Same person, different images
• Negative: Different person (LFW dataset)

🧹 Preprocessing

• Resize: 100x100
• Normalize: [0, 1]
• Augmentation: Brightness, Contrast, Flip

Workflow Diagram

---

📊 Eigenface Analysis (PCA)

• Principal Component Analysis applied to face crops
• Reduced dimensionality while preserving key identity features
• Eigenfaces used as compact input to the deep model

Eigenface Output

---

⚙️ Model Architecture

🧬 Siamese Network with FaceNet Embeddings

Computer-vision-