Research map / 3D vision / Depth and stereo
Learning sRGB-to-Raw-RGB De-Rendering With Content-Aware Metadata
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 content-aware sampling instead of random sampling (ablation) | Raw reconstruction with metadata · NUS patches, three cameras averaged | PSNR: 49.68 -> 50.64 | Helped |
| Use content-aware sampling instead of uniform sampling, with fine-tuning (ablation) | Raw reconstruction with metadata · NUS patches, three cameras averaged | PSNR: 52.59 -> 53.32 | Helped |
| Add online fine-tuning | Ours, raw reconstruction · NUS, Samsung NX2000 | PSNR: 48.08 -> 49.57 | Helped |
| Add online fine-tuning | Ours, raw reconstruction · NUS, Olympus E-PL6 | PSNR: 50.71 -> 51.54 | Helped |
| Add online fine-tuning | Ours, raw reconstruction · NUS, Sony SLT-A57 | PSNR: 50.49 -> 53.11 | Helped |
| Use superpixel max-pooling instead of free-form max-pooling (ablation) | Learned sampler · NUS patches, three cameras averaged | PSNR: 47.75 -> 50.15 | Helped |
| Use both sRGB and RAW images rather than sRGB alone in the sampler (ablation) | Learned sampler · NUS patches, three cameras averaged | PSNR: 49.10 -> 50.15 | Helped |
| Use both sRGB and RAW images rather than RAW alone in the sampler (ablation) | Learned sampler · NUS patches, three cameras averaged | PSNR: 49.72 -> 50.15 | Helped |
| Add the meta loss (ablation) | Reconstruction network · NUS patches, three cameras averaged | PSNR: 50.15 -> 50.64 | Helped |
| Add the meta loss (ablation) | Reconstruction network after fine-tuning · NUS patches, three cameras averaged | PSNR: 53.07 -> 53.32 | Helped |
| Use the learned superpixel sampler instead of SLIC (ablation) | Ours, raw reconstruction · NUS, Samsung NX2000 | PSNR: 45.94 -> 49.57 | Helped |
| Increase the metadata sampling rate from 0% to 0.4% (ablation) | Ours + fine-tuning · NUS, Samsung NX2000 | PSNR: 38.86 -> 48.56 | Helped |
Closest papers by meaning
- Leveraging Frame Affinity for sRGB-to-RAW Video De-rendering (CVPR 2024 · similarity 0.91)
- ReRAW: RGB-to-RAW Image Reconstruction via Stratified Sampling for Efficient Object Detection on the Edge (CVPR 2025 · similarity 0.87)
- DiffRAW: Leveraging Diffusion Model to Generate DSLR-Comparable Perceptual Quality sRGB from Smartphone RAW Images (AAAI 2024 · similarity 0.86)
- RGB-Only Supervised Camera Parameter Optimization in Dynamic Scenes (NeurIPS 2025 · similarity 0.86)
- ISPDiffuser: Learning RAW-to-sRGB Mappings with Texture-Aware Diffusion Models and Histogram-Guided Color Consistency (AAAI 2025 · similarity 0.86)
- RGB-to-Polarization Estimation: A New Task and Benchmark Study (NeurIPS 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.