Research map / Image generation / Image restoration
GIQE: Generic Image Quality Enhancement via Nth Order Iterative Degradation
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 Nth-order iterative degradation augmentation, increasing its order from 1 to 5 (ablation) | GIQE trained for motion deblurring · GOPRO-II (r) and GOPRO-III | PSNR: +1.47db for GOPRO-II(r) at N=1; +1.7db for GOPRO-III at N=1; increasing degradation order further improves performance | Helped |
| Use Nth-order iterative degradation augmentation (ablation) | GIQE trained for motion deblurring · GOPRO-I | PSNR: -0.01db | Hurt |
| Use average pooling for spatial compression instead of convolution (ablation) | GIQE · GOPRO-II (r) | PSNR: -0.28db | Hurt |
| Add spatial compression before attention (ablation) | GIQE · GOPRO-I, GOPRO-II (r), and GOPRO-III | PSNR: +0.18db, +0.22db for pooling and convolution-based SC, respectively | Helped |
| Replace the feed-forward network with multiscale feature extraction using convolution sizes 3 and 5 (ablation) | GIQE · GOPRO-I | PSNR: +0.25db, +0.33db, and +0.55db for size 3, 5, and both respectively | Helped |
| Add the auxiliary degradation-mask decoder (ablation) | GIQE · GOPRO-I, GOPRO-II (r), and GOPRO-III | PSNR: +0.57db, +0.15db, and +0.07db | Helped |
| Add multiscale branches and merge their features with the feature selection module (ablation) | GIQE · GOPRO-I | PSNR: +0.34db and +0.76db as we increase the number of scaled versions to 3 | Helped |
| Sharpen the ground-truth images with randomized Gaussian blur (ablation) | GIQE · GOPRO-I, GOPRO-II (r), and GOPRO-III | PSNR: +0.26db, +0.14db, and +0.08db | Helped |
| Replace Swin Transformer with CSWin Transformer (ablation) | Multiscale GIQE · GOPRO-I and GOPRO-III | PSNR: 33.05 / 25.07 -> 33.21 / 25.19 | Helped |
| Train for 100 additional epochs (ablation) | GIQE · GOPRO-I, GOPRO-II (r), and GOPRO-III | PSNR: 33.05 -> 33.37, 27.05 -> 27.11, and 25.07 -> 25.42 | Helped |
| Use GIQE instead of Uformer | Deblurring models · GoPRO | PSNR: 32.27 -> 33.48 | Helped |
| Use GIQE instead of Uformer | Deblurring models pretrained on GoPRO · RealBlur | PSNR: 30.74 -> 30.18 | Hurt |
Closest papers by meaning
- GenDeg: Diffusion-based Degradation Synthesis for Generalizable All-In-One Image Restoration (CVPR 2025 · similarity 0.89)
- MODEM: A Morton-Order Degradation Estimation Mechanism for Adverse Weather Image Recovery (NeurIPS 2025 · similarity 0.89)
- Unsupervised Degradation Representation Aware Transform for Real-World Blind Image Super-Resolution (AAAI 2025 · similarity 0.87)
- Continuous Adverse Weather Removal via Degradation-Aware Distillation (CVPR 2025 · similarity 0.87)
- AWRaCLe: All-Weather Image Restoration Using Visual In-Context Learning (AAAI 2025 · similarity 0.86)
- Towards Universal Rainy Image Restoration: Benchmark and Baseline (AAAI 2025 · similarity 0.86)
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.