Research map / Image generation / Super-resolution
Learning Local Implicit Fourier Representation for Image Warping
Jaewon Lee, Kwang Pyo Choi, Kyong Hwan Jin
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 amplitude estimator (ablation) | LTEW · DIV2KW homography in-scale | mPSNR: 30.77 -> 30.83 | Improved |
| add long skip connection (ablation) | LTEW · DIV2KW homography in-scale | mPSNR: 30.79 -> 30.83 | Improved |
| use 256 instead of 128 estimated frequencies (ablation) | LTEW · DIV2KW homography in-scale | mPSNR: 30.80 -> 30.83 | Improved |
| use shape-dependent phase (Jacobian and Hessian) instead of none (ablation) | LTEW · DIV2KW homography in-scale | mPSNR: 30.69 -> 30.83 | Improved |
| use shape-dependent phase (Jacobian and Hessian) instead of none (ablation) | LTEW · DIV2KW homography out-of-scale | mPSNR: 26.76 -> 26.73 | No clear effect |
| use shape terms instead of spatially-invariant cell (ablation) | LTEW · DIV2KW homography in-scale | mPSNR: 25.23 -> 30.69 | Improved |
| use shape terms instead of spatially-invariant cell (ablation) | LTEW · DIV2KW homography out-of-scale | mPSNR: 25.91 -> 26.76 | Improved |
| use Fourier features (LTEW) instead of concatenation (LIIF) (ablation) | LTEW · DIV2KW homography in-scale | mPSNR: 30.65 -> 30.83 | Improved |
| use Fourier features (LTEW) instead of transformer (ITSRN) (ablation) | LTEW · DIV2KW homography in-scale | mPSNR: 30.64 -> 30.83 | Improved |
| replace ReLU decoder with sinusoidal activation (ablation) | LTEW · DIV2KW homography in-scale | mPSNR: 30.83 -> 30.80 | Hurt |
| add Jacobian and Hessian shape equation to LIIF (ablation) | LIIF · DIV2KW homography in-scale | mPSNR: 30.65 -> 30.74 | Improved |
| use LTEW instead of Arb-RCAN for out-of-scale asymmetric SR | RCAN · Set5 (x3/x8) out-of-scale | PSNR: 28.37 -> 29.26 | Improved |
Closest papers by meaning
- Towards Progressive Multi-Frequency Representation for Image Warping (CVPR 2024 · similarity 0.90)
- Local Texture Estimator for Implicit Representation Function (CVPR 2022 · similarity 0.89)
- Neural Implicit Morphing of Face Images (CVPR 2024 · similarity 0.86)
- Neural FFTs for Universal Texture Image Synthesis (NeurIPS 2020 · similarity 0.86)
- Learning Spatially Collaged Fourier Bases for Implicit Neural Representation (AAAI 2024 · similarity 0.85)
- How Learnable Grids Recover Fine Detail in Low Dimensions: A Neural Tangent Kernel Analysis of Multigrid Parametric Encodings (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.
