Research map / Image generation / Image restoration
Image Restoration for Under-Display Camera
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 the real-paired-data UNet instead of DeP | UDC restoration · T-OLED | PSNR/SSIM: 28.50/0.9117 -> 36.71/0.9713 | Helped |
| use the real-paired-data UNet instead of DeP | UDC restoration · P-OLED | PSNR/SSIM: 16.97/0.7084 -> 30.45/0.9427 | Helped |
| train UNet on real paired data rather than synthetic data | UNet · T-OLED | PSNR/SSIM: 32.42/0.9343 -> 36.71/0.9713 | Helped |
| train UNet on real paired data rather than synthetic data | UNet · P-OLED | PSNR/SSIM: 25.88/0.9006 -> 30.45/0.9427 | Helped |
| use UNet rather than ResNet | UDC restoration · T-OLED | PSNR/SSIM: 36.26/0.9703 -> 36.71/0.9713 | Helped |
| use UNet rather than ResNet | UDC restoration · P-OLED | PSNR/SSIM: 27.42/0.9176 -> 30.45/0.9427 | Helped |
| double the width of each UNet layer (ablation) | UNet · T-OLED | PSNR/SSIM: 36.71/0.9713 -> 37.00/0.9730 | Helped |
| double the width of each UNet layer (ablation) | UNet · P-OLED | PSNR/SSIM: 30.45/0.9427 -> 30.37/0.9425 | Hurt |
| use two sub-encoders instead of a single encoder (ablation) | UNet · T-OLED | PSNR/SSIM: 36.47/0.9704 -> 36.71/0.9713 | Helped |
| use two sub-encoders instead of a single encoder (ablation) | UNet · P-OLED | PSNR/SSIM: 30.26/0.9387 -> 30.45/0.9427 | Helped |
| add SSIM loss to L1 loss (ablation) | UNet · UDC test set, both display types | PSNR/SSIM and LPIPS: 36.71/0.9713 -> 36.69/0.9714; 0.1209 -> 0.1246; 30.45/0.9427 -> 30.37/0.9403; 0.2219 -> 0.2131 | Mixed |
| add VGG perceptual loss to L1 loss (ablation) | UNet · UDC test set, both display types | PSNR/SSIM and LPIPS: 36.71/0.9713 -> 36.31/0.9711; 0.1209 -> 0.1130; 30.45/0.9427 -> 30.37/0.9403; 0.2219 -> 0.2130 | Mixed |
Closest papers by meaning
- Decoupling Degradations with Recurrent Network for Video Restoration in Under-Display Camera (AAAI 2024 · similarity 0.90)
- Bricker to BRACE: A Bracket Exposure RAW Dataset and Restoration Model for Flicker-Banding (NeurIPS 2026 · similarity 0.81)
- RIFLE: Removal of Image Flicker-Banding via Latent Diffusion Enhancement (NeurIPS 2026 · similarity 0.81)
- SIDL: A Real-World Dataset for Restoring Smartphone Images with Dirty Lenses (AAAI 2025 · similarity 0.80)
- Visual-Instructed Degradation Diffusion for All-in-One Image Restoration (CVPR 2025 · similarity 0.80)
- Learning to Remove Wrinkled Transparent Film with Polarized Prior (CVPR 2024 · similarity 0.80)
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.