A PyTorch implementation of SRGAN specific for Anime Super Resolution based on "Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network". And another PyTorch WGAN-gp implementation of SRGAN referring to "Improved Training of Wasserstein GANs".

This repo contains the project work carried out for the course Deep Learning in my B. Tech Final Year DA-IICT. It is the replication of the code in simpler terms available on GitHub.

The demand for high resolution is common in computer vision applications when it comes to performance in analysis and pattern recognition. However, high-resolution images are not always available.In this work, we evaluate the case of SR using a GAN.

A novel framework for Single Image Super Resolution using Generative Adversarial Network (GAN), namely "Wide Activation with Enhanced Perception Super Resolution GAN (WAEP SRGAN)".

This repository leverages Generative Adversarial Networks (GANs) to enhance image resolution for various applications, using the Super-Resolution GAN (SRGAN) architecture. The project includes a Jupyter Notebook for model training and a detailed research paper documenting the methodology and results.

No description available

Super resolution of single images using SRGANs. Demonstration of super-resolution on single images using SRGANs, built and fine-tuned for this task, done on CelebA dataset. Showcased the enhancement in the image resolution on multiple trials.

No description available

single-image super-resolution using GANs (SRGAN)

Medical X-ray image enhancement using Super-Resolution GAN (SRGAN) to improve image quality for better diagnosis.

This project implements SRGAN (Super-Resolution GAN) using PyTorch, a deep learning model that takes low-resolution (LR) images and generates corresponding high-resolution (HR) images.

SRGAN (Super-Resolution GAN) enhances low-resolution images into high-resolution, realistic outputs using adversarial and perceptual loss. This implementation is trained on the DIV2K dataset.

Super Resolution GAN (SRGAN) is a deep learning model that enhances low-resolution images into high-resolution ones. It uses a generator to create detailed images and a discriminator to make them look more realistic.

MSc dissertation: SRGAN (Super-Resolution GAN) enhances low-res smartphone skin images for improved cancer classification using ResNet-50. Achieved 91.7% accuracy on the PAD-UFES-20 smartphone dataset. 🧬

Welcome! This project is all about single-image super-resolution for satellite images, using a GAN-based approach (SRGAN) and the OLI2MSI dataset. If you're working with remote sensing data and want to boost the resolution of your imagery, you're in the right place.

SRGAN (Super-Resolution Generative Adversarial Network) 🚀 enhances low-res images by generating high-res versions with realistic details. 🖼️ Introduced in 2017, it uses GANs and perceptual loss to outperform traditional methods. Ideal for image restoration, video, and photography. 🎥📸