Research map / Image generation / Image restoration
Attention-Based Adaptive Selection of Operations for Image Restoration in the Presence of Unknown Combined Distortions
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 input-conditioned attention to select among parallel operations (ablation) | Operation-wise attention network · Combined distortions, Mix | PSNR: 23.24 -> 24.71 | Helped |
| Use input-conditioned rather than fixed learned operation weights (ablation) | Operation-wise attention network · Mix | PSNR: 23.37 -> 24.71 | Helped |
| Use input-conditioned attention rather than no attention (ablation) | Operation-wise attention network · Raindrop | PSNR: 26.93 -> 28.24 | Helped |
| Use input-conditioned attention rather than no attention (ablation) | Operation-wise attention network · Blur | PSNR: 21.74 -> 23.66 | Helped |
| Use input-conditioned attention rather than fixed learned weights (ablation) | Operation-wise attention network · Noise | PSNR: 30.58 -> 31.93 | Helped |
| Use input-conditioned attention rather than fixed learned weights (ablation) | Operation-wise attention network · JPEG | PSNR: 29.20 -> 30.07 | Helped |
| Restore combined distortions with operation-wise attention instead of DnCNN | DIV2K mild (unseen) · DIV2K mild (unseen) | PSNR: 27.51 -> 28.33 | Helped |
| Restore combined distortions with operation-wise attention instead of DnCNN | DIV2K moderate · DIV2K moderate | PSNR: 26.50 -> 27.07 | Helped |
| Restore combined distortions with operation-wise attention instead of DnCNN | DIV2K severe (unseen) · DIV2K severe (unseen) | PSNR: 25.26 -> 25.88 | Helped |
| Restore distorted images with operation-wise attention before object detection instead of RL-Restore | SSD300 · PASCAL VOC2007 val | mAP: 66.0 -> 69.3 | Helped |
| Restore distorted images with operation-wise attention before object detection instead of RL-Restore | SSD300 · PASCAL VOC2012 val | mAP: 62.0 -> 67.3 | Helped |
| Train the proposed model on combined-distortion rather than single-distortion data | Operation-wise attention network · Gaussian noise removal | PSNR: 35.20 -> 31.49 | Hurt |
Closest papers by meaning
- Dual Residual Networks Leveraging the Potential of Paired Operations for Image Restoration (CVPR 2019 · similarity 0.88)
- MR-VNet: Media Restoration using Volterra Networks (CVPR 2024 · similarity 0.86)
- A Restoration Network as an Implicit Prior (ICLR 2024 · similarity 0.86)
- Adapt or Perish: Adaptive Sparse Transformer with Attentive Feature Refinement for Image Restoration (CVPR 2024 · similarity 0.85)
- Boosting Image Restoration via Priors from Pre-trained Models (CVPR 2024 · similarity 0.85)
- Omni-Kernel Network for Image Restoration (AAAI 2024 · similarity 0.85)
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.