Research map / Architectures and efficiency / CNN design
Convolutional Neural Network Architecture for Geometric Matching
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 |
|---|---|---|---|
| Add an affine transformation ensemble before the thin-plate spline stage (ablation) | Geometric matching CNN · Proposal Flow | PCK (%): 56 -> 57 | Helped |
| Add the thin-plate spline refinement after affine alignment (ablation) | Geometric matching CNN · Proposal Flow | PCK (%): 49 -> 56 | Helped |
| Use the learned affine estimator instead of RANSAC with the same VGG-16 pool4 features | Affine matching · Proposal Flow | PCK (%): 47 -> 49 | Helped |
| Normalize the correlation map (ablation) | Geometric matching CNN · Proposal Flow | PCK (%): 44 -> 49 | Helped |
| Replace correlation matching with feature concatenation (ablation) | Geometric matching CNN · Proposal Flow | PCK (%): 49 -> 26 | Hurt |
| Replace correlation matching with feature subtraction (ablation) | Geometric matching CNN · Proposal Flow | PCK (%): 49 -> 18 | Hurt |
| Train the affine network on Pascal-synth-aff instead of StreetView-synth-aff (ablation) | Geometric matching CNN · Proposal Flow | PCK (%): 49 -> 45 | No clear effect |
| Train on StreetView-synth-aff rather than Pascal-synth-aff (ablation) | Geometric matching CNN · Proposal Flow | PCK (%): 45 -> 49 | Helped |
Closest papers by meaning
- Convolutional Hough Matching Networks (CVPR 2021 · similarity 0.91)
- Deep Semantic Feature Matching (CVPR 2017 · similarity 0.88)
- Universal Correspondence Network (NeurIPS 2016 · similarity 0.87)
- MATCHA: Towards Matching Anything (CVPR 2025 · similarity 0.87)
- ETO:Efficient Transformer-based Local Feature Matching by Organizing Multiple Homography Hypotheses (NeurIPS 2024 · similarity 0.86)
- Learning Affine Correspondences by Integrating Geometric Constraints (CVPR 2025 · similarity 0.86)
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.