Research map / Image generation / Image restoration
DeepLPF: Deep Local Parametric Filters for Image Enhancement
Sean Moran, Pierre Marza, Steven McDonagh, Sarah Parisot, Gregory Slabaugh
Check 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 |
|---|---|---|---|
| Add elliptical filters (ablation) | U-Net image enhancement · MIT-Adobe-5K-DPE | PSNR: 21.57 -> 22.56 | Improved |
| Add graduated filters (ablation) | U-Net image enhancement · MIT-Adobe-5K-DPE | PSNR: 21.57 -> 22.64 | Improved |
| Add a Cubic-10 polynomial filter (ablation) | U-Net image enhancement · MIT-Adobe-5K-DPE | PSNR: 21.57 -> 22.69 | Improved |
| Replace Cubic-10 with Cubic-20 (ablation) | U-Net image enhancement · MIT-Adobe-5K-DPE | PSNR: 22.69 -> 23.44 | Improved |
| Combine elliptical and graduated filters with Cubic-20 (ablation) | U-Net+Cubic-20 · MIT-Adobe-5K-DPE | PSNR: 23.44 -> 23.93 | Improved |
| Increase the number of graduated and elliptical filters per channel (ablation) | DeepLPF · MIT-Adobe-5K-DPE | PSNR | Improved |
| Use DeepLPF instead of DPE | Image enhancement · MIT-Adobe-5K-DPE | PSNR: 23.80 -> 23.93 | Improved |
| Use DeepLPF instead of DeepUPE | Image enhancement · MIT-Adobe-5K-UPE | SSIM: 0.893 -> 0.887 | Hurt |
| Use DeepLPF instead of DeepUPE | Image enhancement · MIT-Adobe-5K-UPE | PSNR: 23.04 -> 24.48 | Improved |
| Use DeepLPF instead of U-Net | RAW-to-RGB enhancement · SID Fuji | PSNR: 26.61 -> 26.82 | Improved |
Later papers that compared with this method
7 later accepted papers report DeepLPF as a baseline in their results tables; 7 of them report a better number on most of the results they share with it.
- Learned Image Enhancement via Color Naming (ECCV 2024 · better on 3 of 3 shared results)
- Fast Context-Based Low-Light Image Enhancement via Neural Implicit Representations (ECCV 2024 · better on 1 of 1 shared result)
- Neural Color Operators for Sequential Image Retouching (ECCV 2022 · better on 1 of 1 shared result)
- StarEnhancer: Learning Real-Time and Style-Aware Image Enhancement (ICCV 2021 · better on 3 of 3 shared results)
- Real-Time Image Enhancer via Learnable Spatial-Aware 3D Lookup Tables (ICCV 2021 · better on 2 of 2 shared results)
- Adaptive Unfolding Total Variation Network for Low-Light Image Enhancement (ICCV 2021 · better on 1 of 1 shared result)
- Image Harmonization With Transformer (ICCV 2021 · better on 1 of 1 shared result)
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
- Underexposed Photo Enhancement Using Deep Illumination Estimation (CVPR 2019 · similarity 0.87)
- StarEnhancer: Learning Real-Time and Style-Aware Image Enhancement (ICCV 2021 · similarity 0.86)
- Deep Parametric 3D Filters for Joint Video Denoising and Illumination Enhancement in Video Super Resolution (AAAI 2023 · similarity 0.86)
- Normalize Filters! Classical Wisdom for Deep Vision (NeurIPS 2025 · similarity 0.85)
- AGLLDiff: Guiding Diffusion Models Towards Unsupervised Training-free Real-world Low-light Image Enhancement (AAAI 2025 · similarity 0.84)
- Paint by Inpaint: Learning to Add Image Objects by Removing Them First (CVPR 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.
