AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

A Spatial RNN Codec for End-to-End Image Compression

Chaoyi Lin, Jiabao Yao, Fangdong Chen, Li Wang

CVPR 2020 · Architectures and efficiency › RNNs and normalization · approach: Differentiable optimization layers · 34 citations (OpenAlex)

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use block-based recurrent processing instead of a fully convolutional network (ablation)Spatial RNN Codec · Kodakrate-distortion performanceImproved
use HRNN to pass information across adjacent blocks (ablation)Spatial RNN Codec · Kodakrate-distortion performanceImproved
use HRNN instead of parallelizable BRNN (ablation)Spatial RNN Codec · RGB Kodakencoding and decoding time: 8.719 -> 27.214 ms encoding; 7.495 -> 34.549 ms decodingHurt
reduce BRNN block size from 128×128 to 64×64 (ablation)Spatial RNN Codec · Kodakrate-distortion performanceHurt
reduce BRNN block size from 128×128 to 64×64 (ablation)Spatial RNN Codec · RGB Kodakencoding and decoding time: 8.719 -> 10.304 ms encoding; 7.495 -> 11.355 ms decodingHurt
add adaptive quantization and joint hyperpriors instead of a single hyperprior without adaptive quantization (ablation)Spatial RNN Codec · Kodakrate-distortion performanceMixed
use the Spatial RNN Codec instead of HM16.14image compression · KodakBD-Rate: -26.73%Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy