Research map / Image generation / Fast diffusion sampling
Learnable Fractional Reaction-Diffusion Dynamics for Under-Display ToF Imaging and Beyond
Xin Qiao, Matteo Poggi, Xing Wei, Pengchao Deng, Yanhui Zhou, Stefano Mattoccia
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 fractional calculus (fractional-order dynamics) (ablation) | LFRD2 · RUD-ToF | MAE: 17.00 -> 16.73 | Improved |
| add continuous convolution (ablation) | LFRD2 · RUD-ToF | MAE: 16.88 -> 16.73 | Improved |
| use proposed continuous convolution instead of NFC (ablation) | LFRD2 · RUD-ToF | MAE: 16.97 -> 16.73 | Improved |
| use variable learned fractional order instead of fixed order 0.1 (ablation) | LFRD2 · RUD-ToF | MAE: 18.12 -> 17.62 | Improved |
| use variable learned fractional order instead of fixed order 0.7 (ablation) | LFRD2 · RUD-ToF | MAE: 19.01 -> 17.62 | Improved |
| drop depth from continuous convolution inputs (ablation) | LFRD2 · RUD-ToF | MAE: 16.91 -> 16.73 | Improved |
| add LFRD2 on PE-ToF as initial state builder | PE-ToF · RUD-ToF | MAE / RMSE: 21.2 / 48.7 -> 20.0 / 33.7 | Improved |
| add LFRD2 on UD-ToFnet as initial state builder | UD-ToFnet · RUD-ToF | MAE / RMSE: 17.3 / 31.1 -> 16.7 / 30.9 | Improved |
| add GRU iterative refinement instead of LFRD2 | UD-ToFnet · RUD-ToF | MAE / RMSE: 17.29 -> 17.02; 31.11 -> 31.09 | Mixed |
| add LSTM iterative refinement instead of LFRD2 | UD-ToFnet · RUD-ToF | MAE / RMSE: 17.29 -> 16.96; 31.11 -> 31.22 | Mixed |
| add NLSPN diffusion refinement | UD-ToFnet · RUD-ToF | MAE: 17.29 -> 17.23 | No clear effect |
| apply LFRD2 to depth super-resolution at 16x | DSR-EI · NYUv2 | MSE: 57.0 -> 52.3 | Improved |
Closest papers by meaning
- Relay Diffusion: Unifying diffusion process across resolutions for image synthesis (ICLR 2024 · similarity 0.84)
- FAM Diffusion: Frequency and Attention Modulation for High-Resolution Image Generation with Stable Diffusion (CVPR 2025 · similarity 0.84)
- Simpler Diffusion: 1.5 FID on ImageNet512 with Pixel-space Diffusion (CVPR 2025 · similarity 0.84)
- RaPD: Resolution-Agnostic Pixel Diffusion via Semantics-Enriched Implicit Representations (NeurIPS 2026 · similarity 0.83)
- iToF-flow-based High Frame Rate Depth Imaging (CVPR 2024 · similarity 0.83)
- Generating Content for HDR Deghosting from Frequency View (CVPR 2024 · similarity 0.83)
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.
