/Generative-AI

This project utilizes PyTorch to create realistic synthetic images with a Generative Adversarial Network (GAN). Adversarial training refines the Generator model for better image quality and the saved models allow easy image generation.

Primary LanguageJupyter Notebook

Generative Adversarial Networks (GAN) for Synthetic Image Generation

This repository contains a Generative Adversarial Network (GAN) implemented in TensorFlow using PyTorch. The GAN is capable of generating synthetic images for any given dataset. The architecture consists of a discriminator model responsible for distinguishing between real and fake images, and a generator model tasked with generating realistic-looking synthetic images.

Training

You can use your own dataset for training the GAN. Ensure the dataset is preprocessed, and images are converted into tensors and normalized.

Evaluation

After training, you can visualize both real and generated images to assess the training progress qualitatively.

Saved Models

The trained discriminator and generator models are saved for future use, allowing you to generate synthetic images without retraining.

Generating Images

Using the pretrained models, you can generate synthetic images by running the appropriate script.