DRASIC: Distributed Recurrent Autoencoder for Scalable Image Compression

[DCC 2020] This is an implementation of DRASIC: Distributed Recurrent Autoencoder for Scalable Image Compression

• Slepian-Wolf Achievable Region

• Deep Scalable Distributed Source Coding

Requirements

• Python 3
• PyTorch 1.0

Results

• Rate-distortion curves for data sources distributed by random subsets with T = 16 for all sources.

• Rate-distortion curves for data sources distributed by class labels with T = 16 for all sources.

• Rate-distortion curves for data sources distributed by random subsets with T=16 for the first half of sources and $T=8$ for the second half of sources.

• Rate-distortion curves for data sources distributed by class labels with T=16 for the first half of sources and $T=8$ for the second half of sources.

Acknowledgement

Enmao Diao
Jie Ding
Vahid Tarokh