Conditional Face Generation using GAN + VAE

This project combines Generative Adversarial Networks (GAN) and Variational Autoencoders (VAE) to generate high-quality human face images based on conditions such as age, gender, and hair attributes.

🔍 Overview

• Model: GAN + VAE (hybrid architecture)
• Dataset: CelebA Dataset (>200,000 labeled face images)
• Technologies: Python, TensorFlow, NumPy, Google Colab
• Result: Generated face images are clear and match specified conditions.

📊 Sample Results

📌 Notes

• This is a personal research project (2024–2025).
• The model can be extended to support more attributes or higher resolution.
• GitHub repository is currently being updated.

---

Author: Nguyễn Lương Duy – duy.nl204960@sis.hust.edu.vn