Research map / Image generation / Editing and style transfer
Supervised Homography Learning with Realistic Dataset Generation
Hai Jiang, Haipeng Li, Songchen Han, Haoqiang Fan, Bing Zeng, Shuaicheng Liu
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 |
|---|---|---|---|
| generate training pairs with realistic motion (dominant plane + real non-dominant region) instead of single-image warping | DHN · CA-unsup test | PME (AVG): 1.37 -> 0.34 | Improved |
| retrain supervised networks on generated CA-sup dataset | IHN · CA-unsup test | PME (AVG): 4.67 -> 1.19 | Improved |
| retrain supervised networks on generated CA-sup dataset | DHN · GHOF test | PME (AVG): 6.61 -> 3.01 | Improved |
| add content consistency module (ablation) | RealSH · CA-unsup test | PME (AVG): 0.37 -> 0.34 | Improved |
| add quality assessment module (ablation) | RealSH · CA-unsup test | PME (AVG): 0.35 -> 0.34 | Improved |
| add CCM and QAM on top of G-phase-only baseline (ablation) | RealSH · CA-unsup test | PME (AVG): 0.38 -> 0.34 | Improved |
| increase iterations from 2 to 3 (ablation) | RealSH · CA-unsup test | PME (AVG): 0.34 -> 0.33 | No clear effect |
| increase iterations from 1 to 2 (ablation) | RealSH · CA-unsup test | PME (AVG): 0.38 -> 0.34 | Improved |
| use CAHomo as backbone with the framework (ablation) | RealSH · CA-unsup test | PME (AVG): 0.88 -> 0.44 | Improved |
| use HomoGAN as backbone with the framework (ablation) | RealSH · CA-unsup test | PME (AVG): 0.39 -> 0.30 | Improved |
| supervised training on generated data versus unsupervised training (ablation) | BasesHomo · CA-unsup test | PME (AVG): 0.50 -> 0.38 | Improved |
| compare with feature-based method in regular scenes | RealSH · CA-unsup test (RE) | PME: 0.22 vs 0.22 (HomoGAN); 0.30 (SIFT+RANSAC) | No clear effect |
Closest papers by meaning
- Adapting Dense Matching for Homography Estimation with Grid-based Acceleration (CVPR 2025 · similarity 0.85)
- MCNet: Rethinking the Core Ingredients for Accurate and Efficient Homography Estimation (CVPR 2024 · similarity 0.85)
- Mask-Homo: Pseudo Plane Mask-Guided Unsupervised Multi-Homography Estimation (AAAI 2024 · similarity 0.84)
- HomoGen: Enhanced Video Inpainting via Homography Propagation and Diffusion (CVPR 2025 · similarity 0.84)
- SSHNet: Unsupervised Cross-modal Homography Estimation via Problem Reformulation and Split Optimization (CVPR 2025 · similarity 0.84)
- Noisy One-point Homographies are Surprisingly Good (CVPR 2024 · 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.
