Research map / Probabilistic and scientific modeling / Differential equations and PDEs
Learning Physics-based Forward Model Corrections in Unrolled Networks for Diffuser-based Imaging
Yoko Sogabe, Shiori Sugimoto, Shoichiro Saito, Masaki Kitahara
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 Eigen-Kernel forward model in the last 2 iterations (ablation) | EK-DUN · Real lensless test set | PSNR (dB): 29.54 -> 31.09 | Improved |
| add Eigen-Kernel forward model in all 10 iterations instead of last 2 (ablation) | EK-DUN · Real lensless test set | PSNR (dB): 31.09 -> 31.12 | No clear effect |
| add Eigen-Kernel forward model in all 10 iterations instead of last 2 (ablation) | EK-DUN · Real lensless test set | runtime (s): 0.187 -> 0.246 | Hurt |
| add Eigen-Kernel in last 2 iterations to K-base only (ablation) | EK-DUN · Real lensless test set | runtime (s): 0.181 -> 0.187 | Hurt |
| use EK-DUN instead of Le-ADMM-U | EK-DUN · Real lensless test set | PSNR (dB): 27.65 -> 31.09 | Improved |
| use EK-DUN instead of UPDN | EK-DUN · Real lensless test set | PSNR (dB): 30.65 -> 31.09 | Improved |
| use EK-DUN instead of UPDN | EK-DUN · Real lensless test set | runtime (s): 0.257 -> 0.187 | Improved |
| use EK-DUN instead of UPDN | EK-DUN · Real lensless test set | peak GPU memory (GiB): 2.86 -> 2.03 | Improved |
| use EK-DUN instead of UPDN | EK-DUN · Real lensless test set | LPIPS: 0.2764 -> 0.2650 | Improved |
| use EK-DUN instead of UPDN | EK-DUN · Real lensless test set | trainable params (M): 11.9 -> 82.7 | Hurt |
Closest papers by meaning
- Integrated Forward–Inverse Network for Reconstruction for Lensless Image Reconstruction (ECCV 2026 · similarity 0.88)
- Video Diffusion Posterior Sampling for Seeing Beyond Dynamic Scattering Layers (TPAMI 2025 · similarity 0.86)
- A Physics-Informed Blur Learning Framework for Imaging Systems (CVPR 2025 · similarity 0.85)
- Light Transport-aware Diffusion Posterior Sampling for Single-View Reconstruction of 3D Volumes (CVPR 2025 · similarity 0.84)
- A Physics-informed Low-rank Deep Neural Network for Blind and Universal Lens Aberration Correction (CVPR 2024 · similarity 0.84)
- LENS: Low-Frequency Eigen Noise Shaping for Efficient Diffusion Sampling (NeurIPS 2026 · 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.
