Research map / Graph learning / Graph algorithms and clustering
Deep Graph Matching via Blackbox Differentiation of Combinatorial Solvers
Michal Rolínek, Paul Swoboda, Dominik Zietlow, Anselm Paulus, Vít Musil, Georg Martius
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 |
|---|---|---|---|
| add global-feature-conditioned affinity weights (ablation) | BB-GM · Pascal VOC | keypoint matching accuracy (%): 77.9 -> 80.1 | Improved |
| fine-tune the VGG backbone (ablation) | BB-GM · Pascal VOC | keypoint matching accuracy (%): 78.9 -> 80.1 | Improved |
| add quadratic (edge) costs to unary costs (ablation) | BB-GM · Pascal VOC | keypoint matching accuracy (%): 78.6 -> 80.1 | Improved |
| use blackbox combinatorial solver instead of Sinkhorn (ablation) | BB-GM · Pascal VOC | keypoint matching accuracy (%): 77.9 -> 80.1 | Improved |
| allow partial matchings instead of forcing maximal matchings (ablation) | BB-GM · Pascal VOC unfiltered keypoints | F1 score (%): 51.9 -> 61.4 | Improved |
| add multi-graph matching post-processing with 5-image sets | BB-GM · Pascal VOC unfiltered keypoints | F1 score (%): 61.4 -> 62.8 | Improved |
| replace DGMC with BB-GM | DGMC* · Pascal VOC (intersection filtering) | keypoint matching accuracy (%): 73.2 -> 80.1 | Improved |
| replace DGMC with BB-GM | DGMC* · SPair-71k hard viewpoint | keypoint matching accuracy (%): 61.3 -> 70.6 | Improved |
| switch from intersection to inclusion test filtering (ablation) | DGMC · Pascal VOC | keypoint matching accuracy (%): 73.2 -> 73.1 | No clear effect |
| replace DGMC with BB-GM, pretrain only | DGMC* · Willow ObjectClass motorbike | keypoint matching accuracy (%): 69.8 -> 95.8 | Improved |
| replace DGMC with BB-GM, Pascal pretrain and Willow train | DGMC* · Willow ObjectClass car | keypoint matching accuracy (%): 96.5 -> 95.7 | No clear effect |
Closest papers by meaning
- Learning Combinatorial Solver for Graph Matching (CVPR 2020 · similarity 0.92)
- Deep Learning of Graph Matching (CVPR 2018 · similarity 0.92)
- Learning Combinatorial Embedding Networks for Deep Graph Matching (ICCV 2019 · similarity 0.91)
- Combinatorial Learning of Robust Deep Graph Matching: An Embedding Based Approach (TPAMI 2020 · similarity 0.91)
- Learning Partial Graph Matching via Optimal Partial Transport (ICLR 2025 · similarity 0.91)
- Discrete Cycle-Consistency Based Unsupervised Deep Graph Matching (AAAI 2024 · similarity 0.91)
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.
