AICT: An Adaptive Image Compression Transformer

TL;DR

AICT introduces an adaptive image compression transformer that captures global and local contexts more effectively, leading to improved compression efficiency and image quality compared to existing methods.

Contribution

It proposes a novel Transformer-based auto-regressive prior model and a learnable scaling module with ConvNeXt components for better latent representation and image reconstruction.

Findings

Outperforms VVC reference encoder in efficiency

Achieves higher quality image reconstruction

Enhances global and local context modeling

Abstract

Motivated by the efficiency investigation of the Tranformer-based transform coding framework, namely SwinT-ChARM, we propose to enhance the latter, as first, with a more straightforward yet effective Tranformer-based channel-wise auto-regressive prior model, resulting in an absolute image compression transformer (ICT). Current methods that still rely on ConvNet-based entropy coding are limited in long-range modeling dependencies due to their local connectivity and an increasing number of architectural biases and priors. On the contrary, the proposed ICT can capture both global and local contexts from the latent representations and better parameterize the distribution of the quantized latents. Further, we leverage a learnable scaling module with a sandwich ConvNeXt-based pre/post-processor to accurately extract more compact latent representation while reconstructing higher-quality images. Extensive experimental results on benchmark datasets showed that the proposed adaptive image compression transformer (AICT) framework significantly improves the trade-off between coding efficiency and decoder complexity over the versatile video coding (VVC) reference encoder (VTM-18.0) and the neural codec SwinT-ChARM.