Research map / 3D vision / Human pose estimation
IMP: Iterative Matching and Pose Estimation With Adaptive Pooling
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 |
|---|---|---|---|
| Use iterative geometry-aware matching and pose estimation (IMP) | SuperPoint matcher · YFCC100m | AUC @5°: 37.1 -> 39.4 | Helped |
| Use efficient iterative matching with adaptive pooling (EIMP) | SuperPoint matcher · YFCC100m | AUC @5°: 37.1 -> 37.9 | Helped |
| Use IMP instead of SGMNet | SuperPoint matcher · Scannet | AUC @5°: 16.4 -> 16.6 | Helped |
| Use EIMP instead of SGMNet | SuperPoint matcher · Scannet | AUC @5°: 16.4 -> 15.9 | Hurt |
| Add shared attention (ablation) | IMP · YFCC100m, SuperPoint | AUC @5°: 36.9 -> 38.6 | Helped |
| Add pose-consistency loss (ablation) | IMP-S · YFCC100m, SuperPoint | AUC @5°: 38.6 -> 39.1 | Helped |
| Add pose-aware iteration (ablation) | IMP-SC · YFCC100m, SuperPoint | AUC @5°: 39.1 -> 39.4 | Helped |
| Add adaptive pooling (ablation) | EIMP-S · YFCC100m, SuperPoint | AUC @5°: 35.4 -> 36.4 | Helped |
| Add pose-consistency loss (ablation) | EIMP-SA · YFCC100m, SuperPoint | AUC @5°: 36.4 -> 36.9 | Helped |
| Add pose-aware iteration (ablation) | EIMP-SCA · YFCC100m, SuperPoint | AUC @5°: 36.9 -> 37.2 | Helped |
| Use pose uncertainty to adjust adaptive sampling (ablation) | EIMP-SCPA · YFCC100m, SuperPoint | AUC @5°: 37.2 -> 37.9 | Helped |
| Replace adaptive pooling with ratio-based sampling (ablation) | EIMP · YFCC100m, SuperPoint | Mean matching score (M.S.): 18.4 -> 7.8 | Hurt |
Closest papers by meaning
- RUBIK: A Structured Benchmark for Image Matching across Geometric Challenges (CVPR 2025 · similarity 0.86)
- Learning Affine Correspondences by Integrating Geometric Constraints (CVPR 2025 · similarity 0.85)
- SecondPose: SE(3)-Consistent Dual-Stream Feature Fusion for Category-Level Pose Estimation (CVPR 2024 · similarity 0.85)
- Unifying Correspondence Pose and NeRF for Generalized Pose-Free Novel View Synthesis (CVPR 2024 · similarity 0.85)
- RefPose: Leveraging Reference Geometric Correspondences for Accurate 6D Pose Estimation of Unseen Objects (CVPR 2025 · similarity 0.85)
- From Correspondences to Pose: Non-minimal Certifiably Optimal Relative Pose without Disambiguation (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.