Research map / Detection and segmentation / CNN features
Localization in the Crowd with Topological Constraints
Shahira Abousamra, Minh Hoai, Dimitris Samaras, Chao Chen
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 the persistence loss to DICE training (ablation) | TopoCount · ShanghaiTech A | GAME G(3): 114 -> 104 | Improved |
| Replace weighted BCE with DICE plus persistence loss (ablation) | TopoCount · ShanghaiTech A | GAME G(3): 122 -> 104 | Improved |
| Set persistence-loss weight λ to 1.5 rather than 1.0 (ablation) | TopoCount · ShanghaiTech A | GAME G(3): 104 -> 104 | No clear effect |
| Set persistence-loss weight λ to 2.0 rather than 1.0 (ablation) | TopoCount · ShanghaiTech A | GAME G(3): 104 -> 107 | Hurt |
| Use 50×50 instead of 150×150 persistence-loss patches (ablation) | TopoCount · ShanghaiTech A | GAME G(3): 114.2 -> 104.9 | Improved |
| Use 50×50 instead of 30×30 persistence-loss patches (ablation) | TopoCount · ShanghaiTech A | GAME G(3): 108.2 -> 104.9 | Improved |
| Use 100×100 instead of 30×30 persistence-loss patches (ablation) | TopoCount · UCF-QNRF | GAME G(3): 222.1 -> 206.6 | Improved |
| Use 100×100 instead of 150×150 persistence-loss patches (ablation) | TopoCount · UCF-QNRF | GAME G(3): 208.5 -> 206.6 | Improved |
| Concatenate TopoCount maps as extra inputs to CAN | CAN · ShanghaiTech A | MAE: 62.3 -> 59.5 | Improved |
| Concatenate TopoCount maps as extra inputs to Bayesian | Bayesian · ShanghaiTech A | MAE: 62.8 -> 58 | Improved |
| Concatenate TopoCount maps as extra inputs to CAN | CAN · ShanghaiTech B | RMSE: 12.2 -> 13.2 | Hurt |
| Concatenate TopoCount maps as extra inputs to Bayesian | Bayesian · JHU++ validation | MAE: 59.3 -> 54.1 | Improved |
Closest papers by meaning
- Rethinking Counting and Localization in Crowds: A Purely Point-Based Framework (ICCV 2021 · similarity 0.88)
- Optimal Transport Minimization: Crowd Localization on Density Maps for Semi-Supervised Counting (CVPR 2023 · similarity 0.87)
- Exploring Scale Shift in Crowd Localization under the Context of Domain Generalization (IJCV 2026 · similarity 0.87)
- Point-to-Region Loss for Semi-Supervised Point-Based Crowd Counting (CVPR 2025 · similarity 0.84)
- Towards Scalable Topological Regularizers (ICLR 2025 · similarity 0.83)
- Recognition through Reasoning: Reinforcing Image Geo-localization with Large Vision-Language Models (NeurIPS 2025 · similarity 0.83)
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.
