Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network

shreya chawla (~shreya13)


Super-resolution Generative Adversarial Networks is a type of GAN which can enhance the resolution/quality of images. Enhancing the quality of images has many use-cases like:

  1. To recover old low-resolution images
  2. To automatically enhance the quality of the camera feed in video surveillance, images transferred over the Internet and television broadcasting and many more!

In this talk, I plan to cover the following topics:

  1. A brief introduction to GAN
  2. Introduction to the Super Resolution Problem
  3. Introduction to SRGANs and its architecture
  4. Model training in Keras
  5. Using the trained model to enhance the quality of images
  6. Briefly discuss improvements made by ESRGAN (Enhanced SRGAN)

This talk will provide an introduction to SRGANs and training SRGANs. After the talk, attendees will be able to train their own SRGAN network!


  1. Proficiency in the Python programming language
  2. Basics of neural networks
  3. Optional basic understanding of GANs (I will introduce briefly)
  4. Experience working in Keras/Tensorflow 2.0

Speaker Info:

Shreya Chawla is a senior year computer science and engineering graduate student. She has experience working on deep learning, computer vision, machine learning, natural language processing, and speech processing and synthesis. She has interned at CSIR-CDRI labs as a data science intern and is undergoing an internship at IIIT - H as a research intern. As a part of CSI, Cybros, and DSC clubs in her college, she encourages and guides fellow students to use Python language and explore its vastness. She has mentored teams to work on technologies like recommender system and computer vision based Python projects. She actively takes part in hackathons around her. In her free time, she watches series and listens to music.

Speaker Links:

  1. LinkedIn:
  2. Github:

Id: 1425
Section: Data Science, Machine Learning and AI
Type: Talks
Target Audience: Intermediate
Last Updated: