Research map / Image generation / Image restoration
The Conditional Lucas & Kanade Algorithm
Chen-Hsuan Lin, Rui Zhu, Simon Lucey
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 |
|---|---|---|---|
| learn conditional image gradients (Conditional LK) instead of generative gradients (Generative LK) (ablation) | Conditional LK · Multi-PIE planar alignment (affine and homography) | frequency of convergence | Improved |
| use Conditional LK instead of SDM when training data is small | Conditional LK · Multi-PIE planar alignment | frequency of convergence vs number of training samples | Improved |
| use Conditional LK instead of SDM or IC-LK | Conditional LK · Multi-PIE planar alignment | convergence rate (speed) | Improved |
| give SDM more than 100 examples per layer | SDM · Multi-PIE planar alignment | convergence speed: over 100 examples per layer | Mixed |
| swap warp function at test time (train on similarity or homography, test on affine) (ablation) | Conditional LK · Multi-PIE planar alignment, 100 examples per layer | frequency of convergence | No clear effect |
| use Conditional LK instead of IC-LK with dense LBP descriptors | Conditional LK · Multi-PIE alignment with dense LBP (K=8) | frequency of convergence vs initial warp scale | Improved |
| train Conditional LK instead of SDM on LBP features with few samples | Conditional LK · Multi-PIE alignment with dense LBP | frequency of convergence vs training samples: 20 examples per layer vs more than 50 examples per iteration for SDM | Improved |
| use Conditional LK tracker instead of IC-LK | Conditional LK · low frame-rate template tracking with LBP (three videos) | percentage of successfully tracked frames | Improved |
| use Conditional LK instead of IC-LK with a point distribution model | Conditional LK · IJAGS facial model fitting | convergence speed and fitting RMSE | Improved |
Closest papers by meaning
- PRISE: Demystifying Deep Lucas-Kanade With Strongly Star-Convex Constraints for Multimodel Image Alignment (CVPR 2023 · similarity 0.88)
- CLKN: Cascaded Lucas-Kanade Networks for Image Alignment (CVPR 2017 · similarity 0.88)
- Deep Lucas-Kanade Homography for Multimodal Image Alignment (CVPR 2021 · similarity 0.88)
- FastJAM: a Fast Joint Alignment Model for Images (NeurIPS 2025 · similarity 0.86)
- Training-free Dense-Aligned Diffusion Guidance for Modular Conditional Image Synthesis (CVPR 2025 · similarity 0.86)
- Orientation Matters: Making 3D Generative Models Orientation-Aligned (NeurIPS 2025 · similarity 0.85)
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.
