Research map / Detection and segmentation / CNN features
Point in, Box Out: Beyond Counting Persons in Crowds
Check 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 |
|---|---|---|---|
| Add iterative pseudo ground-truth updating (ablation) | PSDDN · SHA | MAE: 168.6 -> 104.7 | Helped |
| Replace classic box regression with locally-constrained regression (ablation) | PSDDN · SHA | MAE: 104.7 -> 89.8 | Helped |
| Add curriculum learning (ablation) | PSDDN · SHA | MAE: 89.8 -> 85.4 | Helped |
| Add iterative pseudo ground-truth updating (ablation) | PSDDN · SHB | MAE: 69.8 -> 41.7 | Helped |
| Replace classic box regression with locally-constrained regression (ablation) | PSDDN · SHB | MAE: 41.7 -> 19.1 | Helped |
| Add curriculum learning (ablation) | PSDDN · SHB | MAE: 19.1 -> 16.1 | Helped |
| Add iterative pseudo ground-truth updating (ablation) | PSDDN · SHA | AP: 0.308 -> 0.491 | Helped |
| Add iterative pseudo ground-truth updating (ablation) | PSDDN · SHB | AP: 0.015 -> 0.241 | Helped |
| Use point supervision with PSDDN rather than fully supervised Faster R-CNN | WiderFace · WiderFace hard | AP: 0.347 -> 0.396 | Helped |
| Use point supervision with PSDDN rather than fully supervised Faster R-CNN | WiderFace · WiderFace easy | AP: 0.840 -> 0.605 | Hurt |
| Use point supervision with PSDDN rather than fully supervised Faster R-CNN | WiderFace · WiderFace medium | AP: 0.724 -> 0.605 | Hurt |
| Use PSDDN instead of Li et al. | TRANCOS · TRANCOS | GAME0: 3.56 -> 4.79 | Hurt |
Closest papers by meaning
- Context-Aware Crowd Counting (CVPR 2019 · similarity 0.91)
- Leveraging Self-Supervision for Cross-Domain Crowd Counting (CVPR 2022 · similarity 0.90)
- Top-Down Feedback for Crowd Counting Convolutional Neural Network (AAAI 2018 · similarity 0.89)
- Shallow Feature Based Dense Attention Network for Crowd Counting (AAAI 2020 · similarity 0.89)
- Divide and Grow: Capturing Huge Diversity in Crowd Images With Incrementally Growing CNN (CVPR 2018 · similarity 0.88)
- Point-to-Region Loss for Semi-Supervised Point-Based Crowd Counting (CVPR 2025 · similarity 0.87)
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.