Research map / Video / Video segmentation and tracking
Temporal Epipolar Regions
Check 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 |
|---|---|---|---|
| Restrict candidate matches to valid Temporal Epipolar Regions (ablation) | Nearest-neighbor SIFT matching · Real datasets | correct matching: 7.1% | Helped |
| Restrict candidate matches to valid Temporal Epipolar Regions with nearest-neighbor initialization (ablation) | SIFT matching with initialized Sq · Real datasets | correct matching: 2.7% | Helped |
| Increase known correspondences from |Sq| = 3 to |Sq| = 4 (ablation) | TER search-space prediction · Synthetic data | simulations with up to 20% valid image area: approximately 27% -> approximately 42% | Helped |
| Increase known correspondences to |Sq| = 7 (ablation) | TER search-space prediction · Synthetic data | simulations with up to 20% valid image area: over 65% | Helped |
| Use more than four known correspondences (ablation) | TER search-space prediction · Real datasets | samples with less than 40% valid image area: over 85% | Helped |
| Add pixel-location noise with standard deviation 3 (ablation) | TER prediction with six cameras and three initial correspondences · Synthetic data | simulations with no valid candidate in at least one image: 0.54% | Hurt |
| Add pixel-location noise with standard deviation 5 (ablation) | TER prediction with six cameras and three initial correspondences · Synthetic data | simulations with no valid candidate in at least one image: 3.34% | Hurt |
| Initialize known correspondences using nearest neighbors instead of ground-truth points (ablation) | TER-assisted matching · Real datasets | correct matching: In 294 of the 496 experiments, at least one point in Sq was incorrectly matched. | Hurt |
| Use no TER prediction (ablation) | Nearest-neighbor SIFT matching · Real datasets | correct matching: 7.1% average improvement with TERs | Mixed |
Closest papers by meaning
- Camera Calibration From Dynamic Silhouettes Using Motion Barcodes (CVPR 2016 · similarity 0.86)
- Optimal Relative Pose With Unknown Correspondences (CVPR 2016 · similarity 0.85)
- Epipolar Transformers (CVPR 2020 · similarity 0.85)
- Know Where You Are From: Event-Based Segmentation via Spatio-Temporal Propagation (AAAI 2025 · similarity 0.84)
- Emergent Temporal Correspondences from Video Diffusion Transformers (NeurIPS 2025 · similarity 0.84)
- TimeTracker: Event-based Continuous Point Tracking for Video Frame Interpolation with Non-linear Motion (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.