Research map / Detection and segmentation / CNN features
Learning Spatial Awareness to Improve Crowd Counting
Zhi-Qi Cheng, Jun-Xiu Li, Qi Dai, Xiao Wu, Alexander G. Hauptmann
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 SPANet (MEP loss) | MCNN · ShanghaiTech Part A | MAE: 110.2 -> 99.7 | Improved |
| add SPANet (MEP loss) | CSRNet · ShanghaiTech Part A | MAE: 68.2 -> 62.4 | Improved |
| add SPANet (MEP loss) | SANet · ShanghaiTech Part A | MAE: 67.0 -> 59.4 | Improved |
| add SPANet (MEP loss) | MCNN · ShanghaiTech Part B | MAE: 26.4 -> 19.1 | Improved |
| add SPANet (MEP loss) | MCNN · UCF_CC_50 | MAE: 377.6 -> 292.5 | Improved |
| add SPANet (MEP loss) | SANet · WorldExpo'10 (avg over scenes) | MAE: 8.2 -> 7.7 | Improved |
| add pairwise ranking loss Lr to L2 (ablation) | MCNN · ShanghaiTech Part A | MAE: 110.2 -> 107.2 | Improved |
| add MEP loss to L2 + Lr (ablation) | MCNN · ShanghaiTech Part A | MAE: 107.2 -> 99.7 | Improved |
| remove L2 from the full objective (ablation) | MCNN · ShanghaiTech Part A | MAE: 99.7 -> 99.3 | No clear effect |
| sample patches from bottom right corner instead of center (ablation) | MCNN · ShanghaiTech Part A | MAE: 101.2 -> 99.7 | Improved |
| pool both patches on the same density map instead of a different one (ablation) | MCNN · ShanghaiTech Part A | MAE: 100.3 -> 99.7 | No clear effect |
| grid-search loss weights instead of fixed (ablation) | MCNN · ShanghaiTech Part A | MAE: 99.7 -> 98.3 | No clear effect |
Closest papers by meaning
- AutoScale: Learning to Scale for Crowd Counting (IJCV 2022 · similarity 0.90)
- Attention Scaling for Crowd Counting (CVPR 2020 · similarity 0.90)
- Context-Aware Crowd Counting (CVPR 2019 · similarity 0.89)
- Composition Loss for Counting, Density Map Estimation and Localization in Dense Crowds (ECCV 2018 · similarity 0.89)
- Rethinking Spatial Invariance of Convolutional Networks for Object Counting (CVPR 2022 · similarity 0.89)
- Top-Down Feedback for Crowd Counting Convolutional Neural Network (AAAI 2018 · similarity 0.89)
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.
