Skip to content

Image alignment using CNN. Used autoencoder architecture, spatial transformer network, Deformation vector field estimation and Fully convolutional Neural Netowork

Notifications You must be signed in to change notification settings

vjayd/Image-Alignment-using-CNN

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 

Repository files navigation

Image-Alignment-using-CNN

Problem statement : Given two images 1. real image ; 2 distorted image. The task is to align the distorted image with respect to real image such that both images attain maximum similarity.

Approach : Dataset i have used in this project in MNIST image dataset. Loss function used is MSE loss.

How to reproduce the results

  1. Download the dataset and store it in the data folder.
  2. Run 10,000 iterations, you can change the hyperparameters if you want.
  3. Select any two random images from the data set as test image
  4. Check the results.

Note: You have to change the logic according to your needs.

Some results:

results

Considering optimizing and support for more datasets ,i will be updating the respository on a weekly basis.

About

Image alignment using CNN. Used autoencoder architecture, spatial transformer network, Deformation vector field estimation and Fully convolutional Neural Netowork

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages