Face Recognition System with Fisherface Method and Support Vector Machine

This project is part of my Bachelor's Degree thesis and was also featured in a paper published by IEEE, though it uses a different dataset. The published paper is available at IEEE.

Project Status: Done

Project Intro/Objective

The purpose of this project is to build a face recognition system using the Fisherface method for dimensionality reduction and a Support Vector Machine (SVM) for classification. This project explores effective facial image processing techniques and provides a robust machine learning model for recognizing faces from images. The system has potential applications in security, identity verification, and user authentication systems.

Methods Used

• Image Processing
• Dimensionality Reduction (Eigen decomposition)
• Machine Learning (Support Vector Machine)
• Data Visualization

Technologies

• Python
• OpenCV
• Matplotlib
• Numpy
• Pandas
• MTCNN
• Patool

Project Description

This project focuses on processing face images from the Labeled Faces in the Wild (LFW) dataset. The system applies the following steps:

1. Face Detection and Cropping: Utilizes MTCNN to detect and crop faces.
2. Preprocessing: Converts images to grayscale and resizes them to a uniform size.
3. Dimensionality Reduction: Computes Fisherfaces through Eigen decomposition for feature extraction.
4. Classification: Implements a Support Vector Machine (SVM) to classify faces into categories based on labels.
5. Visualization: Displays the mean face image, eigenfaces, and reconstructed images from the reduced feature space.

Directory Structure

├── code.ipynb         <- Jupyter Notebook containing the code for data preprocessing, prediction, and visualization.
├── README.md          <- Project documentation and instructions.

Getting Started

1. Clone this repository:

   git clone https://github.com/yourusername/FaceRecognition-Fisherfaces.git

2. Raw Data: The dataset is stored in the Labeled Faces in the Wild Home Focused directory. If the dataset is not included, download it from the LFW Dataset.
3. Data Processing Scripts: The data processing and transformation scripts are located in the main directory of this repository.

Setup Instructions

Follow the steps below to set up the environment and execute the system:

1. Install the required Python packages:

   pip install opencv-python-headless numpy matplotlib pandas mtcnn patool

2. Run the Python notebook to execute the system.

Featured Notebooks/Analysis/Deliverables

• Data Preprocessing, Prediction, and Visualization Code

Results

Face Recognition Model

Metric	Value
Accuracy	88.16%
Precision	89.50%
Recall	82.44%
F1-Score	85.00%

Contributing Members

Team Lead
Syachrul Qolbi Nur Septi

Other Members

• Nurul Gusriani
• Iin Irianingsih
• Herlina Napitupulu
• Intan Nurma Yulita