Topic

Artificial neural network classifier trained by backpropagation for face recognition (Famous48 dataset).

Technology used

• Python 3.11
• Tensorflow, Keras - python machine learning library

Face Recognition with Famous48 Dataset

Overview

This repository contains code for a face recognition system using the Famous48 dataset. The project demonstrates training, evaluation, and inference of a deep learning model to classify faces of celebrities in the dataset. It is intended as a starting point for exploring face recognition techniques and neural network architectures.

Works done

• We first load the dataset from raw text files load_dataset.py
• We tried to train the model on version of AlexNet adapted to size of images (24x24) alexnet.ipynb
• We experimented with LeNet-5 architecture which is designed for small images lenet.ipynb
• We decided to adjust hyperparameters with optuna framework optuna_hyperparams_tunning.ipynb
• Then, we tested the LeNet-5 architecture with found parameters from previous step after_optuna.ipynb

Getting Started

1. Clone the Repository

git clone https://github.com/Pawcharz/face-recognition-famous48.git
cd face-recognition-famous48

Contact

For questions or contributions, please contact:

• GitHub: Pawcharz

---

Enjoy exploring face recognition!