Research map / Architectures and efficiency / CNN design
AutoScale: Learning to Scale for Crowd Counting
Chenfeng Xu, Dingkang Liang, Yongchao Xu, Song Bai, Wei Zhan, Xiang Bai, Masayoshi Tomizuka
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 learning-to-scale (L2S) module (ablation) | AutoScale · ShanghaiTech Part A (dense regions) | MAE: 94.2 -> 71.5 | Improved |
| add the learning-to-scale (L2S) module (ablation) | AutoScale · ShanghaiTech Part B (dense regions) | MAE: 12.1 -> 9.3 | Improved |
| add the learning-to-scale (L2S) module (ablation) | AutoScale · UCF-QNRF (dense regions) | MAE: 87.4 -> 78.0 | Improved |
| add the learning-to-scale (L2S) module (ablation) | AutoScale · JHU-Crowd++ test (dense regions) | MAE: 76.9 -> 67.9 | Improved |
| add the learning-to-scale (L2S) module (ablation) | AutoScale · NWPU-Crowd val (dense regions) | MAE: 81.3 -> 70.7 | Improved |
| add L2S to localization-based baseline (ablation) | localization-based AutoScale · ShanghaiTech Part A (dense regions) | MAE: 72.1 -> 55.1 | Improved |
| add L2S to localization-based baseline (ablation) | localization-based AutoScale · UCF-QNRF (dense regions) | MAE: 138.5 -> 110.5 | Improved |
| add L2S to FPN (regression) | AutoScale · ShanghaiTech Part A | MAE: 66.1 -> 60.5 | Improved |
| add L2S to FPN (regression) | AutoScale · UCF-QNRF | MAE: 92.0 -> 87.5 | Improved |
| add L2S to FPN (regression) | AutoScale · NWPU-Crowd test | MAE: 108.3 -> 94.1 | Improved |
| add L2S to FPN (regression) | AutoScale · JHU-Crowd++ test | MAE: 81.5 -> 76.4 | Improved |
| add L2S to FPN* (localization) | localization-based AutoScale · NWPU-Crowd test | MAE: 143.2 -> 123.9 | Improved |
Closest papers by meaning
- Composition Loss for Counting, Density Map Estimation and Localization in Dense Crowds (ECCV 2018 · similarity 0.91)
- Learn to Scale: Generating Multipolar Normalized Density Maps for Crowd Counting (ICCV 2019 · similarity 0.91)
- Recurrent Attentive Zooming for Joint Crowd Counting and Precise Localization (CVPR 2019 · similarity 0.91)
- Optimal Transport Minimization: Crowd Localization on Density Maps for Semi-Supervised Counting (CVPR 2023 · similarity 0.91)
- Locate, Size and Count: Accurately Resolving People in Dense Crowds via Detection (TPAMI 2020 · similarity 0.91)
- Rethinking Counting and Localization in Crowds: A Purely Point-Based Framework (ICCV 2021 · similarity 0.91)
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.
