Research map / Architectures and efficiency / RNNs and normalization
Checkerboard Context Model for Efficient Learned Image Compression
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 |
|---|---|---|---|
| Replace serial context decoding with checkerboard context and two-pass decoding | Minnen2018 · Kodak | decoding time (ms): 1323.66 -> 29.66 | Helped |
| Replace serial context decoding with checkerboard context and two-pass decoding | Minnen2018 · Tecnick | decoding time (ms): 4977.98 -> 95.46 | Helped |
| Replace serial context decoding with checkerboard context and two-pass decoding | Cheng2020 · Kodak | decoding time (ms): 1395.35 -> 75.23 | Helped |
| Replace serial context decoding with checkerboard context and two-pass decoding | Cheng2020 · Tecnick | decoding time (ms): 5296.16 -> 259.37 | Helped |
| Replace serial context with checkerboard context | Minnen2018 and Cheng2020 · Kodak | rate-distortion performance | Hurt |
| Add checkerboard context to the context-free hyperprior | Minnen2018 · Kodak | BDBR: -17.0% | Helped |
| Add checkerboard context to the context-free hyperprior | Cheng2020 · Kodak | BDBR: -27.4% | Helped |
| Use checkerboard 3×3 context instead of serial 3×3 context (ablation) | Random-mask model · Kodak | BPP: 0.3928 -> 0.3651 | Helped |
| Use checkerboard 5×5 context instead of serial 5×5 context (ablation) | Random-mask model · Kodak | BPP: 0.3817 -> 0.3648 | Helped |
| Expand checkerboard context from 3×3 to 5×5 (ablation) | Random-mask model · Kodak | BPP: 0.3651 -> 0.3648 | No clear effect |
| Refer to all eight surrounding neighbours rather than four checkerboard neighbours (ablation) | Random-mask model · Kodak | BPP: 0.3651 -> 0.3649 | No clear effect |
| Add four adjacent-neighbour references rather than disable context (ablation) | Random-mask model · Kodak | BPP: 0.4332 -> 0.3651 | Helped |
Closest papers by meaning
- MambaIC: State Space Models for High-Performance Learned Image Compression (CVPR 2025 · similarity 0.86)
- Causal Context Adjustment Loss for Learned Image Compression (NeurIPS 2024 · similarity 0.86)
- 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 0.84)
- Robustly overfitting latents for flexible neural image compression (NeurIPS 2024 · similarity 0.84)
- Test-time Adaptation for Image Compression with Distribution Regularization (ICLR 2025 · similarity 0.83)
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.