DCGAN (Deep Convolutional Generative Adversarial Network) custom architecture builder and image synthesizer to specify the architecture of the generator and discriminator, visualize the models, train the GAN, synthesize images, and analyze synthetic imagery losslessly visualized.

This research introduces PotatoGANs, a novel data augmentation technique using GANs to generate synthetic potato disease images, improving model generalization in agricultural disease segmentation

Sketch2Face is a deep learning model that can generate realistic human face images from hand-drawn or computer-generated sketches. This implementation uses a conditional GAN architecture based on the pix2pix model, which has shown excellent results for image-to-image translation tasks.

This repository implements a Generative Adversarial Network (GAN) for generating synthetic signatures. The model includes a generator that creates fake signatures and a discriminator that differentiates real from generated ones. It aims to produce realistic signatures for forgery detection and dataset augmentation.

This repository implements a Variational Autoencoder (VAE) using PyTorch to generate synthetic signatures. The model encodes real signatures into a latent space, samples using the reparameterization trick, and reconstructs them via a decoder. Training optimizes reconstruction loss and KL divergence for smooth latent representations.

This repository contains a PyTorch implementation of a Generative Adversarial Network (GAN) designed to generate synthetic cat images. The model was trained on the CIFAR-10 dataset, specifically filtered to use only cat images.