ICASSP 2024. A Parameterized Generative Adversarial Network Using Cyclic Projection for Explainable Medical Image Classification.

henesys2ellinia (CycleGAN)

The goal of this project is to use CycleGAN, a type of Generative Adversarial Network (GAN), to convert original images to Monet-style paintings. This project was motivated by the desire to explore the capabilities of GANs in image-to-image translation tasks, as well as the potential for using GANs to create artwork in different styles.

Theory, Experiments, and Dataset for our newly proposed Deep Learning method for LLM-driven Cycle Consistency and Semantics Aware Self-Supervised Framework for Unpaired LDR ↔ HDR Image Translation

The "CycleGAN for Horse2Zebra" project uses deep learning to translate horse images into zebra images and vice versa. It involves preparing the dataset, building Generators and Discriminators, employing key loss functions, and using the Adam optimizer. The model is trained and evaluated using tools like Python and PyTorch.

Simple Cycle GAN example

Tensroflow serving served CycelGAN model with Streamlit UI deployed on Google Cloud Run

An introduction to cycle GAN neural networks

Fast Cross-Domain Unsupervised Object detection through Online Style Transfer

Domain adaptation of synthetic document images using neural networks

💻 🐈 Added a self-attention layer to the CycleGAN implementation (PyTorch).

An image manipulation project to perform style transfer using cycleGAN paper.

Translate Brain MRI Images of one type to another

Experimental implementation of HRNet as a generator in CycleGAN

Work done for the Spring 2019 class of DEep Learning for Image REstoration and Synthesis at MVA Master's degree

Code repository for Diploma Thesis "Generative Adversarial Networks for pose and style selection in fashion design applications".

Image denoiser to reduce noise in transmitted documents.

Topology-guided cyclic graph generation using GCNs.

horse2zebra