Research map / Detection and segmentation / CNN features
Adaptive NMS: Refining Pedestrian Detection in a Crowd
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 |
|---|---|---|---|
| Replace greedy-NMS with Adaptive-NMS | Faster R-CNN on CityPersons · CityPersons reasonable val | MR^-2: 14.5 -> 12.9 | Helped |
| Replace greedy-NMS with Adaptive-NMS | RFB Net on CityPersons · CityPersons reasonable val | MR^-2: 13.9 -> 12.7 | Helped |
| Replace greedy-NMS with Adaptive-NMS | FPN on CrowdHuman · CrowdHuman full body val | MR^-2: 52.35 -> 49.73 | Helped |
| Replace greedy-NMS with Adaptive-NMS | RFB Net on CrowdHuman · CrowdHuman full body val | MR^-2: 65.22 -> 63.03 | Helped |
| Add Adaptive-NMS to Faster R-CNN with AggLoss at ×1.3 scale | Faster R-CNN on CityPersons test · CityPersons reasonable test | MR^-2: 11.79 -> 11.40 | Helped |
| Replace greedy-NMS with soft-NMS (linear) (ablation) | Faster R-CNN on CityPersons · CityPersons reasonable val | MR^-2: 14.5 -> 14.2 | Helped |
| Replace greedy-NMS with soft-NMS (linear) (ablation) | RFB Net on CityPersons · CityPersons reasonable val | MR^-2: 13.9 -> 14.2 | Hurt |
| Combine soft-NMS with Adaptive-NMS (ablation) | Faster R-CNN on CityPersons · CityPersons reasonable val | MR^-2: 12.9 -> 14.1 | No clear effect |
| Combine soft-NMS with Adaptive-NMS (ablation) | RFB Net on CityPersons · CityPersons reasonable val | MR^-2: 12.7 -> 14.3 | Hurt |
| Add AggLoss (ablation) | Faster R-CNN on CityPersons with Adaptive-NMS · CityPersons reasonable val | MR^-2: 12.9 -> 11.9 | Helped |
| Increase input scale from ×1 to ×1.3 (ablation) | Faster R-CNN with AggLoss and Adaptive-NMS · CityPersons reasonable val | MR^-2: 11.9 -> 10.8 | Helped |
| Replace greedy-NMS with soft-NMS (ablation) | FPN on CrowdHuman · CrowdHuman full body val | MR^-2: 52.35 -> 51.97 | Helped |
Closest papers by meaning
- NMS by Representative Region: Towards Crowded Pedestrian Detection by Proposal Pairing (CVPR 2020 · similarity 0.90)
- Learning Non-Maximum Suppression (CVPR 2017 · similarity 0.88)
- Detection in Crowded Scenes: One Proposal, Multiple Predictions (CVPR 2020 · similarity 0.87)
- Variational Pedestrian Detection (CVPR 2021 · similarity 0.87)
- Accelerating Non-Maximum Suppression: A Graph Theory Perspective (NeurIPS 2024 · similarity 0.86)
- Repulsion Loss: Detecting Pedestrians in a Crowd (CVPR 2018 · similarity 0.86)
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.