https://github.com/arash-keshavarz/face_generation_gan
Generate new fake images from CelebA dataset
Science Score: 13.0%
This score indicates how likely this project is to be science-related based on various indicators:
-
○CITATION.cff file
-
✓codemeta.json file
Found codemeta.json file -
○.zenodo.json file
-
○DOI references
-
○Academic publication links
-
○Academic email domains
-
○Institutional organization owner
-
○JOSS paper metadata
-
○Scientific vocabulary similarity
Low similarity (4.7%) to scientific vocabulary
Repository
Generate new fake images from CelebA dataset
Basic Info
- Host: GitHub
- Owner: Arash-Keshavarz
- Language: Jupyter Notebook
- Default Branch: main
- Size: 1.04 MB
Statistics
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
- Releases: 0
Metadata Files
README.md
Generation of Fake Celebrity Faces
Project Overview
This project leverages Generative Adversarial Networks (GANs) to generate realistic synthetic images using the CelebA dataset. Our objective is to create new faces that closely resemble those of actual celebrities, showcasing the capabilities of modern GAN architectures.
Dataset
We utilized the CelebA dataset, which contains over 200,000 celebrity images with diverse facial attributes. This dataset was preprocessed to ensure high-quality input for the GAN model.
Methodology
We implemented the Wasserstein GAN with Gradient Penalty (WGAN-GP) to enhance the stability and performance of the training process. The WGAN-GP framework helps in generating more realistic and high-quality images by improving the training dynamics of the GAN.
Results
Original Faces
Below is an example of the original faces from the CelebA dataset:
Generated Faces
Below are the generated faces after 10 epochs of training with WGAN-GP:
Owner
- Name: Arash
- Login: Arash-Keshavarz
- Kind: user
- Repositories: 1
- Profile: https://github.com/Arash-Keshavarz
GitHub Events
Total
- Watch event: 3
Last Year
- Watch event: 3