Research map / 3D vision / Depth and stereo
Neural Maximum a Posteriori Estimation on Unpaired Data for Motion Deblurring
Youjian Zhang, Chaoyue Wang, Dacheng Tao
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 |
|---|---|---|---|
| add NeurMAP unpaired training to DMPHN | DMPHN · Kernel-synthesized blur | PSNR: 23.51 -> 24.44 | Improved |
| add NeurMAP unpaired training to DMPHN | DMPHN · Kernel-synthesized blur | LPIPS: 0.2363 -> 0.1237 | Improved |
| add NeurMAP to MPRNet | MPRNet · GoPro | PSNR: 32.66 -> 32.30 | Hurt |
| add NeurMAP to MPRNet | MPRNet · RealBlur | NIQE: 5.308 -> 4.834 | Improved |
| add NeurMAP to DMPHN | DMPHN · URB | NIQE: 5.201 -> 4.302 | Improved |
| add motion estimation and reblurring plus natural discriminator to a sharp discriminator-only GAN (Net1) (ablation) | DMPHN · Kernel-synthesized blur | PSNR: 18.99 -> 24.44 | Improved |
| add image discriminators to motion estimation plus reblurring (Net2) (ablation) | DMPHN · Kernel-synthesized blur | PSNR: 21.94 -> 24.44 | Improved |
| use natural image discriminator instead of sharp image discriminator (ablation) | DMPHN · Kernel-synthesized blur | PSNR: 23.65 -> 24.44 | Improved |
| raise natural discriminator weight to 0.5 (1:0.5) instead of 1:0.1 (ablation) | DMPHN + NeurMAP · Kernel-synthesized blur | PSNR: 24.44 -> 22.16 | Hurt |
| raise sharp weight to 2 (2:0.1) instead of 1:0.1 (ablation) | DMPHN + NeurMAP · Kernel-synthesized blur | PSNR: 24.44 -> 21.42 | Hurt |
| add TV regularization (ablation) | NeurMAP · kernel-synthesized qualitative | visual artifacts | Improved |
| add reblurring loss for the motion estimation network (ablation) | NeurMAP · kernel-synthesized qualitative | motion map quality | Improved |
Closest papers by meaning
- A Neural Approach to Blind Motion Deblurring (ECCV 2016 · similarity 0.90)
- Non-Blind Deblurring: Handling Kernel Uncertainty With CNNs (CVPR 2018 · similarity 0.90)
- Deep Multi-Scale Convolutional Neural Network for Dynamic Scene Deblurring (CVPR 2017 · similarity 0.89)
- Efficient Video Deblurring Guided by Motion Magnitude (ECCV 2022 · similarity 0.89)
- Parameterized Blur Kernel Prior Learning for Local Motion Deblurring (CVPR 2025 · similarity 0.89)
- DynaMoDe-NeRF: Motion-aware Deblurring Neural Radiance Field for Dynamic Scenes (CVPR 2025 · similarity 0.88)
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.
