CoCliCo: Extremely low bitrate image compression based on CLIP semantic and tiny color map - INRIA - Institut National de Recherche en Informatique et en Automatique Access content directly
Conference Papers Year : 2024

CoCliCo: Extremely low bitrate image compression based on CLIP semantic and tiny color map

Abstract

Coding algorithms are usually designed to pixelwisely reconstruct images, which limits the expected gains in terms of compression. In this work, we introduce a semantic compressed representation for images: CoCliCo. We encode the inputs into a CLIP latent vector and a tiny color map, and we use a conditional diffusion model for reconstruction. When compared to the most recent traditional and generative coders, our approach reaches drastic compression gains while keeping most of the high-level information and a good level of realism.
Fichier principal
Vignette du fichier
PCS_2024-2-1.pdf (9.16 Mo) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-04478601 , version 1 (26-02-2024)

Licence

Attribution

Identifiers

  • HAL Id : hal-04478601 , version 1

Cite

Tom Bachard, Tom Bordin, Thomas Maugey. CoCliCo: Extremely low bitrate image compression based on CLIP semantic and tiny color map. PCS 2024 - Picture Coding Symposium, Jun 2024, Taichung, Taiwan. pp.1-5. ⟨hal-04478601⟩
81 View
37 Download

Share

Gmail Facebook X LinkedIn More