ahmed-habashy / dcgan-on-cifar10

In this script, we use Deep Convolutional Generative Adversarial Networks (DCGANs) to generate new images that resemble CIFAR10 dataset images.

DCGAN on CIFAR10 Dataset

In this script, we use Deep Convolutional Generative Adversarial Networks (DCGANs) to generate new images that resemble CIFAR10 dataset images.

Requirements

To run this script, you need to have the following libraries installed: Keras TensorFlow numpy matplotlib

File Structure

Copy code ├── README.md ├── dcgan_cifar10.py

Usage

To run the script, run the following command: python dcgan_cifar10.py

The script will train the generator model for 200 epochs and generate a plot of the generated images every 10 epochs. The generated images will be saved to a file called generated_plot_e<epoch_number>.png.

Note

The script saves the generator model to the file generator_model.h5 and the discriminator model to the file discriminator_model.h5. If you want to use these models for generating new images, you can load them using the Keras load_model() function.

For example, to load the generator model: from keras.models import load_model #load the generator model model = load_model('generator_model.h5')

Credits The code in this script is based on the example provided in the Keras documentation.