Research map / 3D vision / Gaussian splatting
GMM-IKRS: Gaussian Mixture Models for Interpretable Keypoint Refinement and Scoring
Emanuele Santellani, Martin Zach, Christian Sormann, Mattia Rossi, Andreas Kuhn, Friedrich Fraundorfer
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 |
|---|---|---|---|
| Refine keypoints with warp-based robust GMM (GMM-IKRS) | DISK · HPatches v-set | Homography Accuracy AUC @5px: 0.420 -> 0.444 | Improved |
| Refine keypoints with GMM-IKRS | SuperPoint · IMC stereo test set | mAA@10: 0.282 -> 0.318 | Improved |
| Refine keypoints with GMM-IKRS | Harris · HPatches v-set | Repeatability @3px: 0.553 -> 0.532 (Rep); 0.458 -> 0.504 (Rep MNN) | Mixed |
| Refine keypoints with GMM-IKRS | DoG · HPatches v-set | Repeatability @3px: 0.442 -> 0.517 | Improved |
| Refine keypoints with GMM-IKRS | DISK, SuperPoint, R2D2, MD-Net · IMC multiview | mAA@10: DISK 0.729 -> 0.736 (test) | Improved |
| Increase number of warps from 8 to 20 (ablation) | GMM-IKRS on DISK · HPatches v-set | Repeatability @3px: 0.484 -> 0.499 (time 0.43s -> 0.98s) | Improved |
| Add robust outlier weighting to EM (ablation) | GMM-IKRS on DISK · HPatches v-set | Matching Score @3px: 0.395 -> 0.422 | Improved |
| Fit GMM rather than using KDE maxima only (ablation) | GMM-IKRS on DISK · HPatches v-set | Repeatability @3px: 0.473 -> 0.499 | Improved |
| Keep sub-pixel coordinates instead of rounding to integers (ablation) | GMM-IKRS on DISK · HPatches v-set | Repeatability @3px: 0.494 -> 0.499 | Improved |
| Use keypoint-level GMM aggregation instead of aggregating warped heatmaps (ablation) | DISK · HPatches v-set | Repeatability @3px: 0.488 -> 0.499 | Improved |
| Vary KDE bandwidth h within a reasonable range (ablation) | GMM-IKRS · HPatches | GMM fit | No clear effect |
Closest papers by meaning
- Key-Grid: Unsupervised 3D Keypoints Detection using Grid Heatmap Features (NeurIPS 2024 · similarity 0.87)
- Neural Outlier Rejection for Self-Supervised Keypoint Learning (ICLR 2020 · similarity 0.87)
- Towards Complementary Keypoint Detection via Mixture of Detectors (NeurIPS 2026 · similarity 0.87)
- Back to 3D: Few-Shot 3D Keypoint Detection with Back-Projected 2D Features (CVPR 2024 · similarity 0.86)
- Instance-Adaptive and Geometric-Aware Keypoint Learning for Category-Level 6D Object Pose Estimation (CVPR 2024 · similarity 0.86)
- Meta-Point Learning and Refining for Category-Agnostic Pose Estimation (CVPR 2024 · 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.
