Research map / Optimization / Optimizers
PRISE: Demystifying Deep Lucas-Kanade With Strongly Star-Convex Constraints for Multimodel Image Alignment
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 both strong star-convexity constraints to DeepLK training (ablation) | DeepLK · GoogleEarth | SR at small pixel errors | Helped |
| Use Condition 1 alone instead of both constraints (ablation) | PRISE · GoogleEarth | SR across pixel-error thresholds | Mixed |
| Use Condition 2 alone instead of both constraints (ablation) | PRISE · GoogleEarth | SR across pixel-error thresholds | No clear effect |
| Remove both hinge constraints (ablation) | PRISE · GoogleEarth | training loss | Hurt |
| Use a small hinge-loss weight rho (ablation) | PRISE · GoogleEarth | SR | Hurt |
| Use sufficiently large hinge-loss weight rho (ablation) | PRISE · GoogleEarth | SR | Helped |
| Sample multiple lambda values and select maximum hinge losses instead of sharing a predefined lambda (ablation) | PRISE · training | performance | No clear effect |
| Use a midpoint lambda choice (lambda=0.5) (ablation) | PRISE · GoogleEarth | SR | Helped |
| Use sufficiently large strong-convexity parameter mu (ablation) | PRISE · GoogleEarth | SR | No clear effect |
| Add strong star-convexity constraints | DeepLK · MSCOCO | SR (PE<0.1): 17.16 -> 52.77 $\pm$ 12.45 | Helped |
| Add strong star-convexity constraints | DeepLK · GoogleEarth | SR (PE<0.5): 3.50 -> 25.44 | Helped |
| Add strong star-convexity constraints | DeepLK · GoogleMap | SR (PE<0.5): 2.25 -> 48.13 | Helped |
Closest papers by meaning
- FastJAM: a Fast Joint Alignment Model for Images (NeurIPS 2025 · similarity 0.88)
- 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)
- Unsupervised Homography Estimation on Multimodal Image Pair via Alternating Optimization (NeurIPS 2024 · similarity 0.84)
- PARSAC: Accelerating Robust Multi-Model Fitting with Parallel Sample Consensus (AAAI 2024 · similarity 0.84)
- SSHNet: Unsupervised Cross-modal Homography Estimation via Problem Reformulation and Split Optimization (CVPR 2025 · 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.