Research map / Architectures and efficiency / Pruning and efficient tuning
Exploiting Sample Correlation for Crowd Counting With Multi-Expert Network
Xinyan Liu, Guorong Li, Zhenjun Han, Weigang Zhang, Yifan Yang, Qingming Huang, Nicu Sebe
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 multi-expert strategy | FDC-18 · ShanghaiTech Part A | MAE: 65.4 -> 55.4 | Improved |
| add multi-expert strategy | FDC-18 · ShanghaiTech Part B | MAE: 11.4 -> 6.9 | Improved |
| add multi-expert strategy | FDC-18 · UCF-QNRF | MAE: 93.0 -> 76.2 | Improved |
| add multi-expert strategy | FDC-18 · NWPU-Crowd test | MAE: 119.39 -> 74.7 | Improved |
| add multi-expert strategy | CSRNet · UCF-QNRF | MAE: 98.2 -> 83.1 | Improved |
| add multi-expert strategy | MCNN · UCF-QNRF | MAE: 277 -> 234.1 | Improved |
| add multi-expert strategy | MCNN · ShanghaiTech Part A | MAE: 110.6 -> 94.1 | Improved |
| add multi-expert strategy | CSRNet · ShanghaiTech Part A | MAE: 68.2 -> 60.1 | Improved |
| select unstable samples instead of hard samples for subset Q (ablation) | MFDC (NWPU) · NWPU-Crowd | MAE | Improved |
| change number of clusters (merge smallest or split biggest) away from the adaptive number (ablation) | MFDC · ShanghaiTech Part A | MAE | Hurt |
| use co-fine-tune clustering instead of IG-CNN style divide-and-grow clusters (ablation) | MFDC · ShanghaiTech Part A | label consistency and router accuracy | Improved |
| use multi-expert model trained on NWPU without target training | MFDC-18 vs FDC-18 fine-tuned · ST Part A (transfer) | MAE: 63.1 -> 56.9 | Improved |
Closest papers by meaning
- Crowd Counting With Deep Structured Scale Integration Network (ICCV 2019 · similarity 0.88)
- Weakly-Supervised Crowd Counting Learns from Sorting rather than Locations (ECCV 2020 · similarity 0.88)
- Attention Scaling for Crowd Counting (CVPR 2020 · similarity 0.88)
- Multi-Level Bottom-Top and Top-Bottom Feature Fusion for Crowd Counting (ICCV 2019 · similarity 0.87)
- Hybrid Graph Neural Networks for Crowd Counting (AAAI 2020 · similarity 0.87)
- Free Lunch Enhancements for Multi-modal Crowd Counting (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.
