Contextformer: A Transformer with Spatio-Channel Attention for Context Modeling in Learned Image Compression

A. Burakhan Koyuncu, Han Gao, Atanas Boev, Georgii Gaikov, Elena Alshina, Eckehard Steinbach

Publikation: Beitrag in Buch/Bericht/KonferenzbandKonferenzbeitragBegutachtung

27 Zitate (Scopus)

Abstract

Entropy modeling is a key component for high-performance image compression algorithms. Recent developments in autoregressive context modeling helped learning-based methods to surpass their classical counterparts. However, the performance of those models can be further improved due to the underexploited spatio-channel dependencies in latent space, and the suboptimal implementation of context adaptivity. Inspired by the adaptive characteristics of the transformers, we propose a transformer-based context model, named Contextformer, which generalizes the de facto standard attention mechanism to spatio-channel attention. We replace the context model of a modern compression framework with the Contextformer and test it on the widely used Kodak, CLIC2020, and Tecnick image datasets. Our experimental results show that the proposed model provides up to 11% rate savings compared to the standard Versatile Video Coding (VVC) Test Model (VTM) 16.2, and outperforms various learning-based models in terms of PSNR and MS-SSIM.

OriginalspracheEnglisch
TitelComputer Vision – ECCV 2022 - 17th European Conference, Proceedings
Redakteure/-innenShai Avidan, Gabriel Brostow, Moustapha Cissé, Giovanni Maria Farinella, Tal Hassner
Herausgeber (Verlag)Springer Science and Business Media Deutschland GmbH
Seiten447-463
Seitenumfang17
ISBN (Print)9783031197994
DOIs
PublikationsstatusVeröffentlicht - 2022
Veranstaltung17th European Conference on Computer Vision, ECCV 2022 - Tel Aviv, Israel
Dauer: 23 Okt. 202227 Okt. 2022

Publikationsreihe

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Band13679 LNCS
ISSN (Print)0302-9743
ISSN (elektronisch)1611-3349

Konferenz

Konferenz17th European Conference on Computer Vision, ECCV 2022
Land/GebietIsrael
OrtTel Aviv
Zeitraum23/10/2227/10/22

Fingerprint

Untersuchen Sie die Forschungsthemen von „Contextformer: A Transformer with Spatio-Channel Attention for Context Modeling in Learned Image Compression“. Zusammen bilden sie einen einzigartigen Fingerprint.

Dieses zitieren