Research map / Detection and segmentation / CNN features
Variational Pedestrian Detection
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 |
|---|---|---|---|
| Apply AEVB optimization to FreeAnchor | FreeAnchor · CrowdHuman | MR: 52.8 -> 50.7 | Helped |
| Apply AEVB optimization to FCOS | FCOS · CrowdHuman | MR: 48.3 -> 47.6 | Helped |
| Apply AEVB optimization to FreeAnchor | FreeAnchor · CityPersons R | MR: 14.8% to 13.6% | Helped |
| Apply AEVB optimization to FreeAnchor | FreeAnchor · CityPersons HO | MR: 42.8% to 41.5% | Helped |
| Add IoU loss instead of location L1 loss (ablation) | FreeAnchor · CrowdHuman | MR: 52.83 -> 51.61 | Helped |
| Add AEVB optimization (ablation) | FreeAnchor with IoU loss · CrowdHuman | MR: 51.61 -> 50.67 | Helped |
| Add AEVB optimization (ablation) | FCOS with IoU loss · CrowdHuman | MR: 48.31 -> 47.57 | Helped |
| Set KL factor alpha to 10^-2 rather than 10^-4 or 10^-3 (ablation) | FreeAnchor + AEVB · CrowdHuman | MR: 51.96; 51.27; 50.67 | Helped |
| Set KL factor alpha to 10^-3 rather than 10^-4 or 10^-2 (ablation) | FCOS + AEVB · CrowdHuman | MR: 47.78; 47.57; 48.69 | Mixed |
| Set KL factor alpha to 10^-3 rather than 10^-4 or 10^-2 (ablation) | Faster R-CNN + AEVB · CrowdHuman | MR: 40.93; 40.69; 41.19 | Mixed |
| Increase anchors per pixel from 1 to 2 or 4 (ablation) | FreeAnchor + AEVB · CrowdHuman | MR: 50.67 -> 50.31 -> 49.93 | Helped |
| Increase anchors per pixel from 1 to 4 (ablation) | FCOS + AEVB · CrowdHuman | MR: 47.57 -> 47.49 | Helped |
Closest papers by meaning
- Adaptive NMS: Refining Pedestrian Detection in a Crowd (CVPR 2019 · similarity 0.87)
- Generalizable Pedestrian Detection: The Elephant in the Room (CVPR 2021 · similarity 0.86)
- NMS by Representative Region: Towards Crowded Pedestrian Detection by Proposal Pairing (CVPR 2020 · similarity 0.85)
- Pedestrian Detection With Autoregressive Network Phases (CVPR 2019 · similarity 0.85)
- How Far Are We From Solving Pedestrian Detection? (CVPR 2016 · similarity 0.85)
- Where, What, Whether: Multi-Modal Learning Meets Pedestrian Detection (CVPR 2020 · similarity 0.84)
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.