native-ai / FaceFrontalization

FaceFrontalization

Introduction

This repository contains code for face frontalization using Generative Adversarial Networks (GANs). GANs are used to generate realistic frontal-facing images from non-frontal face images. The models implemented here aim to achieve this transformation effectively.

Folder Structure

├── boarder.py
├── config.py
├── dataset.py
├── models
│   ├── SegNet.py
│   ├── __init__.py
│   ├── discriminator.py
│   └── unet.py
├── train.py
├── utils.py
└── weigths
└── logs

• boarder.py: Contains code for visualization using TensorBoard.
• config.py: Configuration file where model and training parameters can be adjusted.
• dataset.py: Handles data loading and preprocessing.
• models: Directory containing model architectures.
  • SegNet.py: SegNet model architecture.
  • discriminator.py: Discriminator model architecture.
  • unet.py: U-Net model architecture.
• train.py: Script for training the models.
• utils.py: Utility functions used in the project.
• weights: Directory to save trained model weights (tracked using Git LFS).
• logs: Directory to store TensorBoard logs.

Usage

1. Configuration: Adjust the parameters in config.py according to your requirements.

2. Training: Execute train.py to train the GAN models. You can run the training using the following command:

   python train.py

3. TensorBoard Visualization: You can visualize the training progress using TensorBoard by running the following command:

   tensorboard --logdir=logs

   Here are examples of the visualizations obtained from TensorBoard:

   • Real Faces:

   • Generated Faces:

4. Downloading Pre-trained Weights and Logs: Due to GitHub's file size limitations, the weights and logs directories are not included in this repository. However, you can download them from the following Google Drive link:

   Download Weights and Logs

Training Details

• The GAN models are trained for a certain number of iterations specified in config.py.
• Model configuration and training parameters can be modified in config.py.
• The training script train.py orchestrates the training process.

Requirements

• Python 3.x
• Other dependencies as specified in the requirements.txt file.