Skip to content

Enhancing the video quality using super resolution and deep neural networks

Notifications You must be signed in to change notification settings

SINRG-Lab/SR360

Repository files navigation

Super Resolution for Improving Video Quality

A Keras and Tensorflow implementation of super resolution using deep neural networks is proposed. The architecture is shown below.

Network Architecture

Training

  • We divide the video temporal consecutive segments of 1 second duration
  • We downsample the images of each segment and feed both downsampled and original image for training
  • Each video segment has its own model (we called it as a micro-model because it is trained on a little data of 30 images)
  • We overfit the model intentionally as there is no test phase

Prerequisites

  • Python3
  • Tensorflow
  • Keras
  • Jupyter Notebook (optional)

Authors

Pranjal Sahu, Mallesham Dasari

About

Enhancing the video quality using super resolution and deep neural networks

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published