N-O Cool-chic: reconcile fast encoding with lightweight decoding for neural image compression
Théophile Blard, Théo Ladune  1@  , Pierrick Philippe, Xiaoran Jiang  2@  , Olivier Deforges  3@  
1 : Orange Labs [Cesson-Sévigné]
Orange Labs
2 : Institut National des Sciences Appliquées - Rennes
IETR, UMR CNRS 6164
3 : Institut d\'Électronique et des Technologies du numéRique
Université de Nantes, Universite de Rennes 1, Université de Rennes, Institut National des Sciences Appliquées - Rennes, Institut National des Sciences Appliquées, CentraleSupélec : UMR6164, Centre National de la Recherche Scientifique

Overfitted image codecs achieve strong compression performance and low decoder complexity by learning a lightweight decoder for each image. Such codecs include Cool-chic, which presents image coding performance on par with VVC while requiring around 2000 multiplications per decoded pixel. However, the encoding time associated with overfitted codecs may be prohibitively long for real-time applications, posing a challenge to their practical implementation in such scenarios. To address this issue, this paper proposes to decrease the encoding complexity of Cool-chic by bypassing the overfitting procedure and complementing the decoder with an encoder network. The proposed non-overfitted (N-O) Cool-chic, significantly reduces encoding complexity by a factor of 1000 compared to Cool-chic, while maintaining competitive performance.


Personnes connectées : 2 Vie privée
Chargement...