Research map / 3D vision / Depth and stereo
MOWA: Multiple-in-One Image Warping Model
Kang Liao, Zongsheng Yue, Zhonghua Wu, Chen Change Loy
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 4 hierarchical heads with sizes 10-12-14-16 instead of uniform 12 (ablation) | MOWA motion estimation · avg first five tasks | PSNR: 20.38 -> 20.48 | Improved |
| add pixel-level residual flow (ablation) | MOWA motion estimation · avg first five tasks | PSNR: 20.48 -> 20.84 | Improved |
| increase uniform control points to 16x16 (ablation) | MOWA · avg first five tasks | PSNR: 20.42 -> 20.02 | Hurt |
| add hierarchical heads and residual flow to baseline (ablation) | MOWA · avg first five tasks | PSNR: 20.29 -> 20.84 | Improved |
| add image-based task classifier (ablation) | MOWA · avg first five tasks | PSNR: 20.48 -> 20.58 | Improved |
| replace image classifier with point-based classifier (ablation) | MOWA · avg first five tasks | PSNR: 20.58 -> 20.63 | Improved |
| add global pooling features to point classifier (ablation) | MOWA · avg first five tasks | PSNR: 20.63 -> 20.83 | Improved |
| point classifier instead of image classifier, parameters (ablation) | MOWA · classifier cost | Parameters: 3.39M -> 0.0592M | Improved |
| add prompt learning module (ablation) | MOWA · all tasks | PSNR: +0.21 | Improved |
| joint multi-task training vs single-task methods on stitched images | MOWA · Stitched image | PSNR: 21.28 -> 20.72 | Hurt |
| multiple-in-one vs single-task method on rotated images | MOWA · Rotated image | PSNR: 19.89 -> 21.01 | Improved |
| multiple-in-one vs single-task method on fisheye images | MOWA · Fisheye image | PSNR: 21.72 -> 22.25 | Improved |
Closest papers by meaning
- Pixel-Wise Warping for Deep Image Stitching (AAAI 2023 · similarity 0.86)
- Semi-Supervised Coupled Thin-Plate Spline Model for Rotation Correction and Beyond (TPAMI 2024 · similarity 0.86)
- Implicit Warping for Animation with Image Sets (NeurIPS 2022 · similarity 0.85)
- Towards Progressive Multi-Frequency Representation for Image Warping (CVPR 2024 · similarity 0.84)
- Image Warping for Image-to-Image Translation (ECCV 2026 · similarity 0.84)
- Neural Implicit Morphing of Face Images (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.
