Research map / Image generation / Image restoration
Deep Optimal Transport: A Practical Algorithm for Photo-realistic Image Restoration
Theo Joseph Adrai, Guy Ohayon, Michael Elad, Tomer Michaeli
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 |
|---|---|---|---|
| apply latent optimal transport (alpha=0) to pre-trained model | SwinIR · SISR x4 | FID: 2.54 -> 1.39 | Improved |
| apply latent optimal transport (alpha=0) | Swin2SR · SR x4 + JPEG q=10 | FID: 205.00 -> 183.80 | Improved |
| apply latent optimal transport (alpha=0) | ESRGAN · SISR x4 | FID: 1.06 -> 0.89 | Improved |
| apply latent optimal transport (alpha=0) | DDRM · SR x4 + AWGN sigma=50 | FID: 36.44 -> 29.71 | Improved |
| apply latent optimal transport (alpha=0) | NLM denoiser · AWGN sigma=50 | FID: 12.84 -> 14.88 | Hurt |
| apply latent optimal transport (alpha=0) | SwinIR · SISR x4 | PSNR: 28.10 -> 25.48 | Hurt |
| interpolate with alpha=0.9 toward original | SwinIR · SISR x4 | PSNR: 28.10 -> 28.15 | Improved |
| use alpha outside [0,1], e.g. alpha in [1,2] | Swin2SR · SR x4 + JPEG q=10 | FID: 183.80 -> 168.12 | Improved |
| use patch size p=3 or 5 vs p=1 in latent unfolding (ablation) | Latent transport · SISR x4 ESRGAN | PSNR and FID | Improved |
| larger patch sizes 7<=p<=15 (ablation) | Latent transport · SISR x4 ESRGAN | PSNR | Hurt |
| vary training sample count S from 10^2 to 10^5 (ablation) | Latent transport · SISR x4 ESRGAN | all metrics | No clear effect |
| compute distribution parameters from paired images (ablation) | Latent transport · SISR x4 ESRGAN | PSNR and FID | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Restormer — Restormer: Efficient Transformer for High-Resolution Image Restoration (CVPR 2022) · better on 2 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Posterior-Mean Rectified Flow: Towards Minimum MSE Photo-Realistic Image Restoration (ICLR 2025 · similarity 0.92)
- Residual-Conditioned Optimal Transport: Towards Structure-Preserving Unpaired and Paired Image Restoration (ICML 2024 · similarity 0.92)
- A Restoration Network as an Implicit Prior (ICLR 2024 · similarity 0.87)
- Acquire and then Adapt: Squeezing out Text-to-Image Model for Image Restoration (CVPR 2025 · similarity 0.87)
- ReFIR: Grounding Large Restoration Models with Retrieval Augmentation (NeurIPS 2024 · similarity 0.87)
- Boosting Image Restoration via Priors from Pre-trained Models (CVPR 2024 · similarity 0.87)
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.
