Research map / Image generation / Image restoration
Raw-JPEG Adapter: Efficient Raw Image Compression with JPEG
Mahmoud Afifi, Ran Zhang, Michael S Brown
Read the paperCheck your idea against 137,217 papers
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add learned gamma and LuT mapping before JPEG instead of plain JPEG | JPEG · S24 test, quality 100 | PSNR: 46.00 -> 49.04 | Improved |
| add learned adapter instead of fixed 2.2 gamma | JPEG + fixed gamma · S24 test, quality 100 | PSNR: 48.38 -> 49.04 | Improved |
| add OLM pre-processing to JPEG | JPEG · S24 test, quality 100 | PSNR: 46.00 -> 43.65 | Hurt |
| add DCT-domain component to the adapter (ablation) | Raw-JPEG Adapter · S24 test, quality 100 | PSNR: 48.49 -> 48.99 | Improved |
| add color/intensity augmentation (ablation) | Raw-JPEG Adapter (w/ DCT) · S24 test, quality 100 | PSNR: 48.99 -> 49.04 | No clear effect |
| add DCT component when using LIC-TCM codec without tuning (ablation) | Raw-JPEG Adapter · S24 test, LIC-TCM | PSNR: 45.11 -> 45.11 | No clear effect |
| tune the adapter for LIC-TCM (ablation) | Raw-JPEG Adapter (w/ DCT) · S24 test, LIC-TCM | PSNR: 45.11 -> 46.03 | Improved |
| add LuT to learned gamma map (ablation) | Raw-JPEG Adapter · S24 test, quality 75 | PSNR: 42.32 -> 42.50 | Improved |
| add DCT to gamma + LuT (ablation) | Raw-JPEG Adapter · S24 test, quality 75 | PSNR: 42.50 -> 42.70 | Improved |
| add ECA channel attention (ablation) | Raw-JPEG Adapter · S24 test, quality 75 | PSNR: 42.70 -> 42.80 | Improved |
| add differentiable JPEG simulator instead of simple quantization (ablation) | Raw-JPEG Adapter · S24 test, quality 75 | PSNR: 42.80 -> 43.58 | Improved |
| add SSIM loss (ablation) | Raw-JPEG Adapter · S24 test, quality 75 | PSNR: 43.06 -> 43.58 | Improved |
Closest papers by meaning
- RAW Image Reconstruction Using a Self-contained sRGB–JPEG Image with Small Memory Overhead (IJCV 2017 · similarity 0.89)
- RAW Image Reconstruction Using a Self-Contained sRGB-JPEG Image With Only 64 KB Overhead (CVPR 2016 · similarity 0.89)
- Edit-aware RAW reconstruction (CVPR 2026 · similarity 0.88)
- Compress Clean Signal from Noisy Raw Image: A Self-Supervised Approach (ICML 2024 · similarity 0.87)
- Leveraging Frame Affinity for sRGB-to-RAW Video De-rendering (CVPR 2024 · similarity 0.84)
- DCTMamba: Advancing JPEG Image Restoration Through Long-Sequence Modeling and Adaptive Frequency Strategy (AAAI 2025 · similarity 0.84)
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.
