“Neural image compression in a nutshell” by Dr Luis Herranz
Neural image compression is a new paradigm where codecs are modeled as deep neural networks whose parameters are learned from data. There has been increasing interest in this paradigm as a possible competitor to traditional image coding methods based on block-based transform coding.
If you are interested in learning more about neural image compression, do not hesitate to read these articles by our researcher Dr Luis Herranz:
Neural image compression in a nutshell:
- Part 1| Main idea: http://www.lherranz.org/2022/08/24/neural-image-compression-in-a-nutshell-part-1-main-idea/
- Part 2 | Architectures and comparison: http://www.lherranz.org/2022/08/31/neural-image-compression-in-a-nutshell-part-2-architectures-and-comparison/
And, if you want to delve more into the topic, we encourage you to read:
- MAE, SlimCAE and DANICE. Towards practical neural image compression: https://www.lherranz.org/2022/09/17/mae-slimcae-and-danice-towards-practical-neural-image-compression/