https://github.com/jianguoz/dnld-face-generation-p5

Using generative adversarial networks(GAN) to generate new images of faces based on Large-scale CelebFaces Attributes (CelebA) Dataset
https://github.com/jianguoz/dnld-face-generation-p5

Last synced: about 2 months ago
JSON representation

Using generative adversarial networks(GAN) to generate new images of faces based on Large-scale CelebFaces Attributes (CelebA) Dataset

Host: GitHub
URL: https://github.com/jianguoz/dnld-face-generation-p5
Owner: jianguoz
Created: 2017-05-25T02:29:10.000Z (almost 8 years ago)
Default Branch: master
Last Pushed: 2017-05-25T20:37:52.000Z (almost 8 years ago)
Last Synced: 2025-01-16T02:31:53.004Z (3 months ago)
Language: Jupyter Notebook
Homepage:
Size: 7.98 MB
Stars: 1
Watchers: 2
Forks: 0
Open Issues: 0
Metadata Files:
- Readme: README.md

Awesome Lists containing this project

README

# DLND_Face-Generation-P5
**Jianguo Zhang, May 22, 2017(updated on May 25, 2017)**

This project uses generative adversarial networks(GAN) to generate new images of faces using [Large-scale CelebFaces Attributes (CelebA) Dataset](http://mmlab.ie.cuhk.edu.hk/projects/CelebA.html).

The Mnist Dataset is a database of handwritten digits, which includes a training set of 60,000 examples, and a test set of 10,000 examples. The CelebA dataset contains over 200,000 celebrity images with annotations.

Since the celebA dataset is much complex, we first `test` and `tune` our neural network on the [Mnist Dataset](http://yann.lecun.com/exdb/mnist/). Then `use` it on the CelebA dataset to generate face images.

Our network is `a little` similiar with [Deep Convolutional GANs(DCGAN)](https://arxiv.org/pdf/1511.06434.pdf), The DCGAN architecture was first explored last year and has seen impressive results in generating new images, we modify its architecture in this project.

The project uses [TensorFlow==1.0.0](https://www.tensorflow.org/) and [python==3.6](https://www.python.org/downloads/release/python-361/) and runs on [AWS EC2 g2.2xlarge](https://aws.amazon.com/ec2/) GPU device. Please make sure that you install all depencies before run the project, the [requirements](https://github.com/JianguoZhang1994/DNLD_Tv_Script_Generation-P3/blob/master/requirements.txt) incldues minimum requirements of depencies. You should also have a GPU device to accelerate training process.

When you run our project. Make sure the loss of the generator is lower than the loss of the discriminator or close to 0.

### Explore dataset: