Research map / Image generation / Image restoration
Panel-Specific Degradation Representation for Raw Under-Display Camera Image Restoration
Youngjin Oh, Keuntek Lee, Jooyoung Lee, Dae-Hyun Lee, Nam Ik Cho
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 |
|---|---|---|---|
| use raw sensor input instead of RGB input (ablation) | DE-UNET · Own UDC dataset, Axon 30 | PSNR (dB): 23.33 -> 24.20 | Improved |
| use raw input with RGB output instead of raw output (ablation) | DE-UNET · Own UDC dataset, Z-Fold 3 | PSNR (dB): 27.18 -> 27.29 | Improved |
| add channel calibration without triplet representation learning (ablation) | DREUDC · Own UDC dataset, Z-Fold 3 | PSNR (dB): 28.66 -> 28.71 | No clear effect |
| add panel-specific representation learning with triplet loss (ablation) | DREUDC · Own UDC dataset, Axon 30 | PSNR (dB): 26.38 -> 26.66 | Improved |
| add panel-specific representation learning with triplet loss (ablation) | DREUDC · Own UDC dataset, Z-Fold 3 | PSNR (dB): 28.71 -> 29.04 | Improved |
| add DREUDC representation learning to restoration network (ablation) | restoration network · T-OLED | PSNR (dB): 37.37 -> 37.55 | Improved |
| add DREUDC representation learning to restoration network (ablation) | restoration network · P-OLED | PSNR (dB): 30.85 -> 31.17 | Improved |
| use DREUDC instead of BNUDC | UDC restoration · Own UDC dataset, Axon 30 | PSNR (dB): 25.85 -> 26.66 | Improved |
| apply raw-input RAW-to-RGB models AWNet and PyNet to UDC | AWNet, PyNet · Own UDC dataset, Axon 30 | PSNR (dB): 24.20 (DE-UNET RAW-to-RGB) -> 20.98 (AWNet) | Hurt |
Closest papers by meaning
- BNUDC: A Two-Branched Deep Neural Network for Restoring Images From Under-Display Cameras (CVPR 2022 · similarity 0.92)
- Decoupling Degradations with Recurrent Network for Video Restoration in Under-Display Camera (AAAI 2024 · similarity 0.90)
- DDR: Exploiting Deep Degradation Response as Flexible Image Descriptor (NeurIPS 2024 · similarity 0.85)
- Learning Where and What to Restore for Composite Image Restoration (NeurIPS 2026 · similarity 0.84)
- Universal Image Restoration Pre-training via Degradation Classification (ICLR 2025 · similarity 0.84)
- Unsupervised Degradation Representation Aware Transform for Real-World Blind Image Super-Resolution (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.
