Research map / Architectures and efficiency / RNNs and normalization
GLU-Net: Global-Local Universal Network for Dense Flow and Correspondences
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 |
|---|---|---|---|
| combine global correlation at the coarse level with local correlations for refinement (ablation) | dense correspondence estimation · HP | AEPE: 34.64 -> 31.64 | Helped |
| add local correlation refinement to global matching (ablation) | dense correspondence estimation · HP | AEPE: 34.64 -> 31.64 | Helped |
| add the adaptive-resolution strategy to GLOCAL-Net (ablation) | GLOCAL-Net · HP | PCK-1px: 10.23 -> 35.26 | Helped |
| add iterative refinement (ablation) | GLU-Net · HP | PCK-1px: 35.26 -> 36.81 | Helped |
| add cyclic consistency to the global correlation (ablation) | GLU-Net · HP | PCK-1px: 36.81 -> 39.55 | Helped |
| add NC-Net to the global correlation (ablation) | GLU-Net · HP | PCK-1px: 39.55 -> 37.62 | Hurt |
| add NC-Net to the global correlation (ablation) | GLU-Net · KITTI-2012 | AEPE: 3.34 -> 3.80 | Hurt |
| add NC-Net to the global correlation (ablation) | GLU-Net · TSS | PCK: 79.21 -> 82.10 | Helped |
| concatenate L-Net features (ablation) | GLU-Net · TSS | PCK: 79.21 -> 82.76 | Helped |
| add semantic-specific NC-Net and L-Net feature concatenation (ablation) | GLU-Net · TSS | PCK: 79.2 -> 82.8 | Helped |
| use the proposed universal network and shared weights rather than task-specific optical-flow methods | GLU-Net · KITTI-2015 | F1-all: 28.50 -> 37.52 | Hurt |
| train DGC-Net on DPED-CityScape-ADE rather than tokyo | DGC-Net · HP | PCK-1px: 12.00 -> 9.19 | Hurt |
Closest papers by meaning
- Dual-Resolution Correspondence Networks (NeurIPS 2020 · similarity 0.87)
- GOCor: Bringing Globally Optimized Correspondence Volumes into Your Neural Network (NeurIPS 2020 · similarity 0.87)
- FC-GNN: Recovering Reliable and Accurate Correspondences from Interferences (CVPR 2024 · similarity 0.87)
- UFM: A Simple Path towards Unified Dense Correspondence with Flow (NeurIPS 2025 · similarity 0.86)
- Adapting Dense Matching for Homography Estimation with Grid-based Acceleration (CVPR 2025 · similarity 0.86)
- SemAlign3D: Semantic Correspondence between RGB-Images through Aligning 3D Object-Class Representations (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.