Research map / 3D vision / Depth and stereo
EDFFDNet: Towards Accurate and Efficient Unsupervised Multi-Grid Image Registration
Haokai Zhu, Bo Qu, Si-Yuan Cao, Runmin Zhang, Shujie Chen, Bailin Yang, Hui-Liang Shen
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 |
|---|---|---|---|
| use EDFFD instead of TPS deformation (ablation) | EDFFDNet · UDIS-D | PSNR: 25.49 -> 25.93 | Improved |
| use EDFFD instead of B-spline FFD (ablation) | EDFFDNet · UDIS-D | PSNR: 25.95 -> 25.93 | No clear effect |
| use EDFFD instead of B-spline FFD (ablation) | EDFFDNet · UDIS-D | Warp time (ms): 39.1 -> 20.6 | Improved |
| use EDFFD instead of TPS (ablation) | EDFFDNet · UDIS-D | Warp time (ms): 30.5 -> 20.6 | Improved |
| raise influence factor theta from 0.25 to 0.75 (ablation) | EDFFD · UDIS-D | PSNR: 25.30 -> 25.93 | Improved |
| raise theta from 0.75 to 1.50 (ablation) | EDFFD · UDIS-D | PSNR: 25.93 -> 25.81 | Hurt |
| add a second local refinement stage with 18x18 grid (ablation) | EDFFDNet · UDIS-D | PSNR: 25.93 -> 26.49 | Improved |
| enlarge stage-2 grid from 12x12 to 24x24 (ablation) | EDFFDNet-2 · UDIS-D | PSNR: 26.43 -> 26.55 | Improved |
| replace MLP with ASMA (Ng=8) (ablation) | EDFFDNet · UDIS-D | PSNR: 25.87 -> 25.93 | Improved |
| replace MLP with ASMA (Ng=8) (ablation) | EDFFDNet · UDIS-D | Parameters (M): 68.9 -> 23.0 | Improved |
| use ASMA instead of MLP at 4x reduced hidden dimension (ablation) | EDFFDNet · UDIS-D | PSNR: 25.76 -> 25.89 | Improved |
| use progressive correlation (global then local) instead of global everywhere (ablation) | EDFFDNet · UDIS-D | PSNR: 25.54 -> 25.93 | Improved |
Closest papers by meaning
- A Plug-and-Play Image Registration Network (ICLR 2024 · similarity 0.87)
- ERNet: Efficient Non-Rigid Registration Network for Point Sequences (ICCV 2025 · similarity 0.87)
- CARL: A Framework for Equivariant Image Registration (CVPR 2025 · similarity 0.86)
- IIRP-Net: Iterative Inference Residual Pyramid Network for Enhanced Image Registration (CVPR 2024 · similarity 0.84)
- DeMo: Deep Motion Field Consensus with Learnable Kernels for Two-view Correspondence Learning (AAAI 2025 · similarity 0.84)
- Occlusion-aware Non-Rigid Point Cloud Registration via Unsupervised Neural Deformation Correntropy (ICLR 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.
