Research map / Detection and segmentation / CNN features
High-Level Semantic Feature Detection: A New Perspective for 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 |
|---|---|---|---|
| add offset prediction (ablation) | CSP · CityPersons Reasonable | MR−2 (%): 11.4 -> 11.0 | Helped |
| add offset prediction (ablation) | CSP with feature maps downsampled by 16 · Caltech | MR−2 IoU=0.5 (%): 20.27 -> 7.41 | Helped |
| use the center point rather than a top or bottom vertex (ablation) | CSP · Caltech | MR−2 IoU=0.5 (%): 7.75 -> 4.62; 6.52 -> 4.62 | Helped |
| predict height rather than width or height and width (ablation) | CSP scale prediction · Caltech | MR−2 IoU=0.5 (%): 5.31 -> 4.62; 4.73 -> 4.62 | Helped |
| use finer feature maps downsampled by 4 rather than 8 or 16 (ablation) | CSP without offset prediction · Caltech | MR−2 IoU=0.5 (%): 7.00 -> 4.62; 20.27 -> 4.62 | Helped |
| use finer feature maps downsampled by 2 rather than 4 (ablation) | CSP · Caltech | MR−2 IoU=0.75 (%): 36.47 -> 30.08 | Helped |
| combine feature maps φ3, φ4, and φ5 (ablation) | CSP with ResNet-50 · Caltech | MR−2 (%): 5.68 -> 4.62 | Helped |
| add center annotation noise in [0, 4] pixels (ablation) | CSP · Caltech | MR−2 (%): 4.62 -> 5.68 | Hurt |
| add center annotation noise in [0, 8] pixels (ablation) | CSP · Caltech | MR−2 (%): 4.62 -> 8.59 | Hurt |
| use center-and-scale prediction instead of the strongest reported competitor | CSP · CityPersons Reasonable | MR−2 (%): 12.0 -> 11.0 | Helped |
| use center-and-scale prediction instead of ALFNet | CSP · CityPersons Heavy | MR−2 (%): 51.9 -> 49.3 | Helped |
| initialize CSP from CityPersons | CSP · Caltech Reasonable | MR−2 (%): 4.5% -> 3.8% | Helped |
Closest papers by meaning
- VLPD: Context-Aware Pedestrian Detection via Vision-Language Semantic Self-Supervision (CVPR 2023 · similarity 0.89)
- Where, What, Whether: Multi-Modal Learning Meets Pedestrian Detection (CVPR 2020 · similarity 0.89)
- Semantic Channels for Fast Pedestrian Detection (CVPR 2016 · similarity 0.88)
- Roboflow100-VL: A Multi-Domain Object Detection Benchmark for Vision-Language Models (NeurIPS 2025 · similarity 0.85)
- Pedestrian Attribute Recognition: A New Benchmark Dataset and a Large Language Model Augmented Framework (AAAI 2025 · similarity 0.85)
- Selective and Orthogonal Feature Activation for Pedestrian Attribute Recognition (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.