Research map / Image generation / Image restoration
Practical Learned Lossless JPEG Recompression With Multi-Level Cross-Channel Entropy Model in the DCT Domain
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 learned DCT-domain recompression instead of CMIX | JPEG recompression · Kodak | BPP: 1.054 -> 0.965 | Helped |
| Use learned DCT-domain recompression instead of CMIX | JPEG recompression · DIV2K | BPP: 0.931 -> 0.892 | Helped |
| Use learned DCT-domain recompression instead of CMIX | JPEG recompression · CLIC.mobile | BPP: 0.804 -> 0.772 | Helped |
| Use learned DCT-domain recompression instead of CMIX | JPEG recompression · CLIC.pro | BPP: 0.648 -> 0.624 | Helped |
| Add shared and explicit cross-color modeling rather than separate per-component hyperpriors (ablation) | JPEG recompression model · Kodak, QP 75 | BPP: 0.983 -> 0.965 | Helped |
| Add explicit cross-color conditioning to shared hyperprior modeling (ablation) | JPEG recompression model · Kodak, QP 75 | BPP: 0.973 -> 0.965 | Helped |
| Model the color components jointly rather than independently with separate MLCC modules (ablation) | JPEG recompression model · Kodak, QP 75 | BPP: 0.968 -> 0.965 | Helped |
| Add Inner Channel to Outer Channel in MLCC (ablation) | MLCC entropy model · Kodak, QP 75 | BPP: 1.027 -> 0.965 | Helped |
| Add space-to-depth and row splitting to Inner Channel in MLCC (ablation) | MLCC entropy model · Kodak, QP 75 | BPP: 1.012 -> 0.965 | Helped |
| Split rows before columns rather than columns before rows (ablation) | MLCC entropy model · Kodak, QP 75 | BPP: 0.988 -> 0.965 | Helped |
| Use non-uniform rather than uniform 8 channel slices (ablation) | MLCC entropy model · Kodak, QP 75 | BPP: 0.986 -> 0.966 | Helped |
Closest papers by meaning
- Practical Full Resolution Learned Lossless Image Compression (CVPR 2019 · similarity 0.86)
- Improved Lossy Image Compression With Priming and Spatially Adaptive Bit Rates for Recurrent Networks (CVPR 2018 · similarity 0.86)
- JPEG Inspired Deep Learning (ICLR 2025 · similarity 0.84)
- GLIC: General Format Learned Image Compression (AAAI 2025 · similarity 0.84)
- Towards Backward-Compatible Continual Learning of Image Compression (CVPR 2024 · similarity 0.83)
- JDEC: JPEG Decoding via Enhanced Continuous Cosine Coefficients (CVPR 2024 · 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.