In this work, we propose an end-to-end block-based auto-encoder system for image compression. We introduce novel contributions to neural-network based image compression, mainly in achieving binarization simulation, variable bit rates with multiple networks, entropy-friendly representations, inference-stage code optimization and performance-improving normalization layers in the auto-encoder. We evaluate and show the incremental performance increase of each of our contributions.
@article{arxiv.1805.10887,
title = {Block-optimized Variable Bit Rate Neural Image Compression},
author = {Caglar Aytekin and Xingyang Ni and Francesco Cricri and Jani Lainema and Emre Aksu and Miska Hannuksela},
journal= {arXiv preprint arXiv:1805.10887},
year = {2018}
}
Comments
Accepted, Workshop and Challenge on Learned Image Compression (CLIC), CVPR 2018