Research map / 3D vision / Face and shape
Geometry Grounding: Elevating Blind Distortion Correction with 3D Structural Priors
Xiang Li, Weimin Shi, Qichuan Geng, Zhong Zhou
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 structural alignment loss Lsa (ablation) | flow-only rectifier · Wireframe+Places365 test | PSNR: 17.65 -> 24.12 | Improved |
| add structural alignment loss Lsa (ablation) | flow-only rectifier · Wireframe+Places365 test | FID: 145.2 -> 58.4 | Improved |
| add structural integrity constraint Lstruc (ablation) | G2DC · Wireframe+Places365 test | PSNR: 24.12 -> 28.25 | Improved |
| add depth topology constraint Ldepth (ablation) | G2DC · Wireframe+Places365 test | PSNR: 28.25 -> 28.95 | Improved |
| add depth topology constraint Ldepth (ablation) | G2DC · Wireframe+Places365 test | FID: 34.2 -> 25.8 | Improved |
| add intrinsic regularization Lint (ablation) | G2DC · Wireframe+Places365 test | PSNR: 28.95 -> 28.68 | Hurt |
| add intrinsic regularization Lint (ablation) | G2DC · Wireframe+Places365 test | FID: 25.8 -> 19.8 | Improved |
| supervise with VGGT instead of DINOv2 (ablation) | G2DC · Wireframe+Places365 test | Depth-MAE: 0.318 -> 0.098 | Improved |
| supervise with VGGT instead of DINOv2 (ablation) | G2DC · Wireframe+Places365 test | PSNR: 24.61 -> 28.68 | Improved |
| add proposed loss to DeepCalib | DeepCalib · Wireframe+Places365 test | PSNR: 20.84 -> 25.50 | Improved |
| add proposed loss to PCN | PCN · Wireframe+Places365 test | PSNR: 24.56 -> 28.40 | Improved |
| add proposed loss to RDTR | RDTR · Wireframe+Places365 test | PSNR: 27.92 -> 29.95 | Improved |
Closest papers by meaning
- Lifting the Structural Morphing for Wide-Angle Images Rectification: Unified Content and Boundary Modeling (ICCV 2025 · similarity 0.88)
- Blind Geometric Distortion Correction on Images Through Deep Learning (CVPR 2019 · similarity 0.88)
- Unlocking Accurate Geometry in 3DGS via Spatially Varying Affine Rectification (NeurIPS 2026 · similarity 0.88)
- Unlocking the Power of Critical Factors for 3D Visual Geometry Estimation (CVPR 2026 · similarity 0.87)
- AlignDiff: Learning Physically-Grounded Camera Alignment via Diffusion (ICCV 2025 · similarity 0.86)
- LoRA3D: Low-Rank Self-Calibration of 3D Geometric Foundation models (ICLR 2025 · similarity 0.86)
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.
