Research map / Detection and segmentation / CNN features
Where, What, Whether: Multi-Modal Learning Meets 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 |
|---|---|---|---|
| Use the full Where, What, and Whether design instead of Faster R-CNN | Pedestrian detection · Citypersons Heavy | MR^-2: 60.6 -> 18.7 | Helped |
| Add the bird-view Where branch (ablation) | Faster R-CNN pedestrian detector · Citypersons Heavy | MR^-2: 60.6 -> 23.3 | Helped |
| Add depth-based What scale estimation (ablation) | Faster R-CNN pedestrian detector · Citypersons Heavy | MR^-2: 60.6 -> 52.6 | Helped |
| Add the corpus-space Whether branch (ablation) | Faster R-CNN pedestrian detector · Citypersons Heavy | MR^-2: 60.6 -> 30.2 | Helped |
| Add the bird-view Where branch (ablation) | CSP · Citypersons Heavy | MR^-2: 49.3 -> 24.6 | Helped |
| Use the full three-branch model rather than Where+What (ablation) | W^3Net · Citypersons Heavy | MR^-2: 21.7 -> 18.7 | Helped |
| Use bird-view width prediction rather than front-view width prediction (ablation) | W^3Net · Citypersons Heavy | MR^-2: 29.0 -> 18.7 | Helped |
| Generate bird view directly, without real-to-synthetic domain transfer (ablation) | W^3Net · Citypersons Heavy | MR^-2: 18.7 -> 32.1 | Hurt |
| Set bird-view angle and camera height to 60° and 20 (ablation) | W^3Net · Citypersons Heavy | MR^-2: 26.9 -> 18.7 | Helped |
| Use the full W^3Net detector rather than the strongest listed baseline | Pedestrian detection · Caltech H^O | MR^-2: 45.18 -> 28.33 | Helped |
Closest papers by meaning
- High-Level Semantic Feature Detection: A New Perspective for Pedestrian Detection (CVPR 2019 · similarity 0.89)
- What Can Help Pedestrian Detection? (CVPR 2017 · similarity 0.89)
- CityPersons: A Diverse Dataset for Pedestrian Detection (CVPR 2017 · similarity 0.88)
- Occluded Pedestrian Detection Through Guided Attention in CNNs (CVPR 2018 · similarity 0.88)
- Multi-View Pedestrian Occupancy Prediction with a Novel Synthetic Dataset (AAAI 2025 · similarity 0.86)
- A Unified Environmental Network for Pedestrian Trajectory Prediction (AAAI 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.