Research map / Detection and segmentation / CNN features
Boosting Detection in Crowd Analysis via Underutilized Output Features
Shaokai Wu, Fengyu Yang
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 Crowd Hat to PSDNN | PSDNN · UCF-QNRF | MAE: 137.5 -> 75.1 | Improved |
| Add Crowd Hat to PSDNN | PSDNN · UCF-QNRF | F1-m: 67.0 -> 86.2 | Improved |
| Add Crowd Hat to PSDNN | PSDNN · WIDER-Face Hard | AP: 39.6 -> 66.5 | Improved |
| Replace output features with CNN features (ablation) | Crowd Hat on PSDNN · UCF-QNRF | MAE: 75.1 -> 129.3 | Hurt |
| Use all output features instead of box area alone (ablation) | Crowd Hat on PSDNN · UCF-QNRF | MAE: 89.1 -> 75.1 | Improved |
| Use all output features instead of box confidence alone (ablation) | Crowd Hat on PSDNN · UCF-QNRF | MAE: 86.9 -> 75.1 | Improved |
| Use all output features instead of proposal area alone (ablation) | Crowd Hat on PSDNN · UCF-QNRF | MAE: 86.7 -> 75.1 | Improved |
| Use all output features instead of proposal confidence alone (ablation) | Crowd Hat on PSDNN · UCF-QNRF | MAE: 87.4 -> 75.1 | Improved |
| Add 2D compression to uncompressed output features (ablation) | Crowd Hat on PSDNN · UCF-QNRF | MAE: 103.6 -> 82.1 | Improved |
| Add 1D distribution vectors to 2D compression (ablation) | Crowd Hat on PSDNN · UCF-QNRF | MAE: 82.1 -> 75.1 | Improved |
| Add region-adaptive NMS (ablation) | PSDNN · UCF-QNRF | F1-m: 67.0 -> 84.9 | Improved |
| Add decouple-then-align (ablation) | PSDNN · UCF-QNRF | MAE: 137.5 -> 76.2 | Improved |
Closest papers by meaning
- Locate, Size and Count: Accurately Resolving People in Dense Crowds via Detection (TPAMI 2020 · similarity 0.86)
- NWPU-Crowd: A Large-Scale Benchmark for Crowd Counting and Localization (TPAMI 2020 · similarity 0.86)
- Adaptive NMS: Refining Pedestrian Detection in a Crowd (CVPR 2019 · similarity 0.85)
- DecideNet: Counting Varying Density Crowds Through Attention Guided Detection and Density Estimation (CVPR 2018 · similarity 0.85)
- CrowdDiff: Multi-hypothesis Crowd Density Estimation using Diffusion Models (CVPR 2024 · similarity 0.85)
- Robust Zero-Shot Crowd Counting and Localization with Adaptive Resolution SAM (ECCV 2024 · similarity 0.85)
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.
