Research map / Architectures and efficiency / RNNs and normalization
ELIC: Efficient Learned Image Compression With Unevenly Grouped Space-Channel Contextual Adaptive Coding
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 changed | Where | Measured | Effect |
|---|---|---|---|
| Use uneven instead of even 10-slice channel grouping (ablation) | SCCTX channel-conditional coding · Kodak | Param. latency: 13.46 -> 6.45 | Helped |
| Use uneven instead of even 10-slice channel grouping (ablation) | SCCTX channel-conditional coding · CLIC Professional | Param. latency: 73.59 -> 31.88 | Helped |
| Add parallel checkerboard spatial context to even grouping (ablation) | SCCTX channel-conditional coding · Kodak | BD-Rate: -3.81 -> -5.02 | Helped |
| Add parallel checkerboard spatial context to uneven grouping (ablation) | SCCTX channel-conditional coding · Kodak | BD-Rate: -2.78 -> -5.63 | Helped |
| Add parallel checkerboard spatial context to even grouping (ablation) | SCCTX channel-conditional coding · CLIC Professional | BD-Rate: -12.42 -> -14.11 | Helped |
| Add parallel checkerboard spatial context to uneven grouping (ablation) | SCCTX channel-conditional coding · CLIC Professional | BD-Rate: -11.54 -> -14.51 | Helped |
| Use serial instead of parallel checkerboard spatial context with uneven grouping (ablation) | SCCTX channel-conditional coding · Kodak | BD-Rate: -5.63 -> -6.99 | Helped |
| Use serial instead of parallel checkerboard spatial context with uneven grouping (ablation) | SCCTX channel-conditional coding · Kodak | Param. latency: 10.10 -> $>10^3$ | Hurt |
| Use serial instead of parallel checkerboard spatial context with uneven grouping (ablation) | SCCTX channel-conditional coding · CLIC Professional | BD-Rate: -14.51 -> -15.51 | Helped |
| Use serial instead of parallel checkerboard spatial context with uneven grouping (ablation) | SCCTX channel-conditional coding · CLIC Professional | Param. latency: 46.06 -> $>10^3$ | Hurt |
| Replace GDN with one residual bottleneck block (ablation) | Ballé2018 transform · Kodak | BD-Rate: 8.23 -> 5.68 | Helped |
| Replace GDN with three residual bottleneck blocks (ablation) | Ballé2018 transform · Kodak | BD-Rate: 8.23 -> -2.94 | Helped |
Closest papers by meaning
- GLIC: General Format Learned Image Compression (AAAI 2025 · similarity 0.88)
- On Disentangled Training for Nonlinear Transform in Learned Image Compression (ICLR 2025 · similarity 0.86)
- Few-Shot Domain Adaptation for Learned Image Compression (AAAI 2025 · similarity 0.85)
- CALLIC: Content Adaptive Learning for Lossless Image Compression (AAAI 2025 · similarity 0.85)
- Diversify, Contextualize, and Adapt: Efficient Entropy Modeling for Neural Image Codec (NeurIPS 2024 · similarity 0.85)
- Linear Attention Modeling for Learned Image Compression (CVPR 2025 · similarity 0.85)
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.