Research map / Label-efficient and robust learning / Semi-supervised learning
Optimal Transport Minimization: Crowd Localization on Density Maps for Semi-Supervised Counting
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 hard pseudo-label point maps instead of soft density-map pseudo-labels (ablation) | Semi-supervised counting with OT-M · UCF-QNRF, 5% labeled | MAE: 137.17 -> 125.32 | Helped |
| Add confidence weighting to GL for unlabeled data (ablation) | Semi-supervised counting with GL · UCF-QNRF, 5% labeled | MAE: 125.32 -> 120.13 | Helped |
| Add confidence weighting to GL for unlabeled data (ablation) | Semi-supervised counting with GL · UCF-QNRF, 5% labeled | MSE: 214.96 -> 208.87 | Helped |
| Add the gating scheme to generalized loss on labeled data (ablation) | Semi-supervised counting · UCF-QNRF, 5% labeled | MAE: 145.59 -> 144.48 | Helped |
| Add confidence weights on labeled data (ablation) | Semi-supervised counting · UCF-QNRF, 5% labeled | MAE: 144.48 -> 138.52 | Helped |
| Use confidence weighting with L2 loss on unlabeled data (ablation) | Semi-supervised counting with soft pseudo-labels · UCF-QNRF, 5% labeled | MAE: 137.17 -> 135.88 | Helped |
| Use LM instead of OT-M to generate hard pseudo-labels (ablation) | Semi-supervised counting · UCF-QNRF, 5% labeled | MAE: 138.52±10.65 -> 148.53±9.53 | Hurt |
| Use GMM instead of OT-M to generate hard pseudo-labels (ablation) | Semi-supervised counting · UCF-QNRF, 5% labeled | MAE: 126.67±7.41 -> 120.13±7.34 | Hurt |
| Use OT-M localization rather than the strongest listed density-map baseline | Crowd localization on density maps · NWPU-Crowd test | F-measure: 0.685 vs. 0.683 | No clear effect |
| Use OT-M rather than GMM for localization | Crowd localization on UCF-QNRF density maps · UCF-QNRF, GL density map | F-measure: 0.739 -> 0.793 | Helped |
Closest papers by meaning
- Learning to Count via Unbalanced Optimal Transport (AAAI 2021 · similarity 0.92)
- A Generalized Loss Function for Crowd Counting and Localization (CVPR 2021 · similarity 0.91)
- Point-to-Region Loss for Semi-Supervised Point-Based Crowd Counting (CVPR 2025 · similarity 0.89)
- CrowdDiff: Multi-hypothesis Crowd Density Estimation using Diffusion Models (CVPR 2024 · similarity 0.87)
- Single Domain Generalization for Crowd Counting (CVPR 2024 · similarity 0.85)
- Gaussian Density Splatting Network (NeurIPS 2026 · 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.