Skip links to main content
Logo of Geometric Image Processing Laboratory,
Computer Science Department Home
Technion Home Page

Projects Last Projects

High Perceptual Quality Single Image Super Resolution

David-Elone Zana and Odelia Bellaiche Bensegnor

Supervised by Theo Adrai


Nowadays, the metric used to calculate the statistic distance between different pictures is the FID (Fréchet Inception Distance): it uses a pretrained inception network and the W2 divergence to approach the distance between them. We assume that the latent representation of each dataset has a Gaussian distribution. We also assume that the Gaussian distribution is not degenerate: we assume that the covariance matrix is a positive definite matrix: all the principal components are not zeros. To these assumptions, we can add the numerical instability and the impossibility to score a single image.
The limits of this method had woke up in us the will to look for a new metric. Can we find a new solution that won’t need those assumptions and with whom we can score a single image while maintaining numerical stability?

Project Report

Please, see project report.

Final Presentation

Please, see final presentation.

Copyright © 2016 by Geometric Image Processing Lab. All rights reserved.