Variational Image Compression with Convolutional Networks

This repository contains the code and materials for a research project on developing and testing variational methods embedded in a Convolutional Neural Network for image compression. Performances will be evaluated with metrics taken from Information Theory.

References

The project is based on the following papers:

1. Ballé, J.; Minnen, D.; Singh, S.; Jin Wang, S.; Johnston, N. Variational image compression with a scale Hyperprior. ICLR, 2018. URL: https://arxiv.org/pdf/1802.01436.pdf

Authors

The project is being developed by

• Pietro Cappelli, M.Sc. Physics of Data, University of Padova
• Alberto Coppi, M.Sc. Physics of Data, University of Padova
• Nicolò Lai, M.Sc. Physics of Data, University of Padova