DCGAN Implementation

This repository contains a PyTorch implementation of a Deep Convolutional Generative Adversarial Network (DCGAN) for image generation.

Overview

The DCGAN is a type of GAN that uses convolutional and transposed convolutional layers in the discriminator and generator networks, respectively. This architecture has been shown to be effective for generating high-quality images.

The code in this repository trains a DCGAN on the CelebA dataset, which contains images of celebrity faces. The trained generator network can then be used to generate new, realistic-looking face images.

Model Architecture

The model architecture consists of a generator and a discriminator network. The generator takes a noise vector as input and generates a realistic-looking image, while the discriminator tries to distinguish between real and generated images.

The specific architecture of the generator and discriminator can be found in the Generator and Discriminator classes defined in the DCGAN.ipynb file.

Results

After training the DCGAN model, you can generate new face images using the trained generator. The generated images will be saved in the logs/fake directory, and you can visualize them using TensorBoard.

Here is an example of generated face images:

Requirements

- PyTorch
- Torchvision
- Numpy
- Matplotlib
- Jupyter Notebook (or any Python environment