Research map / Detection and segmentation / Open-set and few-shot detection
How Far Are We From Solving 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 |
|---|---|---|---|
| Replace original training annotations with pruned annotations to remove annotation errors (ablation) | RotatedFilters · Caltech validation, 1× training | MR_{-2}^{N}: 33.03 -> 25.91 | Helped |
| Replace original training annotations with new annotations (ablation) | RotatedFilters · Caltech validation, 1× training | MR_{-2}^{N}: 33.03 -> 25.74 | Helped |
| Replace original training annotations with pruned annotations to remove annotation errors (ablation) | ACF · Caltech validation, 1× training | MR_{-2}^{N}: 40.97 -> 35.62 | Helped |
| Align the 9× training portion using a model trained on half of the new annotations (ablation) | RotatedFilters · Caltech-USA test | MR_{-2}^{N}: 17.22 -> 14.54 | Helped |
| Use new 1× annotations and align the 9× training data with new annotations (ablation) | RotatedFilters · Caltech-USA test | MR_{-2}^{N}: 17.22 -> 12.96 | Helped |
| Add VGG rescoring to Rot.Filters proposals (ablation) | Rot.Filters · Caltech-USA test | MR_{-2}^{N}: 17.2% -> 17.8% | Hurt |
| Add VGG rescoring to RotatedFilters-New10× proposals (ablation) | RotatedFilters-New10× · Caltech-USA test | MR_{-2}^{N}: 12.9% -> 11.7% | Helped |
| Add bounding-box regression and a second NMS after VGG rescoring (ablation) | RotatedFilters-New10× + VGG · Caltech-USA test | MR_{-2}^{N}: 11.7% -> 10.0% | Helped |
| Add VGG to the RotatedFilters-New10× detector (ablation) | RotatedFilters-New10× · Caltech-USA test | MR_{-4}^{N}: 22.20 -> 28.37 | Hurt |
| Add bounding-box regression and NMS after VGG (ablation) | RotatedFilters-New10× + VGG · Caltech-USA test | MR_{-4}^{N}: 28.37 -> 20.77 | Helped |
| Ignore overlapping false positives during evaluation (ablation) | Top Caltech detectors · Caltech-USA test | MR_{-4}^{O} | Helped |
| Ignore background false positives during evaluation (ablation) | Top Caltech detectors · Caltech-USA test | MR_{-4}^{O} | Helped |
Closest papers by meaning
- CityPersons: A Diverse Dataset for Pedestrian Detection (CVPR 2017 · similarity 0.90)
- Generalizable Pedestrian Detection: The Elephant in the Room (CVPR 2021 · similarity 0.90)
- MixedPeds: Pedestrian Detection in Unannotated Videos Using Synthetically Generated Human-Agents for Training (AAAI 2018 · similarity 0.89)
- Where, What, Whether: Multi-Modal Learning Meets Pedestrian Detection (CVPR 2020 · similarity 0.88)
- Object Detection Benchmarks are Incomplete: The Role of Label Errors and Annotation Uncertainty (NeurIPS 2026 · similarity 0.87)
- Multispectral Pedestrian Detection with Sparsely Annotated Label (AAAI 2025 · 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.