Research map / Image generation / Image restoration
Matching Pixels Using Co-Occurrence Statistics
Rotal Kat, Roy Jevnisek, Shai Avidan
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 |
|---|---|---|---|
| Replace SSD with co-occurrence statistics while keeping color features | Template matching · Dekel et al. 105 pairs (color features) | AUC: 0.43 -> 0.62 | Improved |
| Replace SSD with co-occurrence statistics while keeping deep features | Template matching · Dekel et al. 105 pairs (deep features) | AUC: 0.55 -> 0.67 | Improved |
| Use deep rather than color features with SSD | Template matching · Dekel et al. 105 pairs | AUC: 0.43 -> 0.55 | Improved |
| Combine color and deep features | CoTM · Oron et al. 270 pairs, mean | AUC: 0.62 -> 0.63 | Improved |
| Use external-set prototypes instead of per-image prototypes (ablation) | CoTM with deep features · Oron et al. 270 pairs, mean | AUC: 0.62 -> 0.56 | No clear effect |
| Use external-set prototypes instead of per-image prototypes (ablation) | CoTM with color features · Oron et al. 270 pairs, mean | AUC: 0.53 -> 0.53 | No clear effect |
| Embed pixel values before template matching instead of applying CoTM directly (ablation) | Template matching · Dekel et al. 105 pairs | AUC: 0.62 -> 0.57 | Hurt |
| Use a 3-dimensional rather than 256-dimensional embedding (ablation) | CoETM · Dekel et al. 105 pairs | AUC: 0.57 -> 0.52 | Hurt |
| Run Lucas-Kanade on co-occurrence-embedded images | Lucas-Kanade registration · Dekel et al. 105 pairs | AUC: 0.47 -> 0.52 | Improved |
| Run KCF on co-occurrence-embedded images | KCF tracker · OTB | AUC: 0.42 -> 0.47 | Improved |
| Use co-occurrence statistics on grayscale pixel values (ablation) | CoTM · Template matching | AUC | Hurt |
Closest papers by meaning
- Template Matching With Deformable Diversity Similarity (CVPR 2017 · similarity 0.85)
- Image Matching from Handcrafted to Deep Features: A Survey (IJCV 2020 · similarity 0.85)
- Best-Buddies Similarity—Robust Template Matching Using Mutual Nearest Neighbors (TPAMI 2017 · similarity 0.85)
- MatchDet: A Collaborative Framework for Image Matching and Object Detection (AAAI 2024 · similarity 0.84)
- Co-op: Correspondence-based Novel Object Pose Estimation (CVPR 2025 · similarity 0.83)
- RUBIK: A Structured Benchmark for Image Matching across Geometric Challenges (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.
