Research map / Label-efficient and robust learning / Recognition and long-tail
Learning Disentangled Attribute Representations for Robust Pedestrian Attribute Recognition
Read the paperCheck 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 shared pooled features with attribute-specific features learned by SSCA (ablation) | DAFL on PA100k · PA100k | mA: 80.71 -> 82.21 | Helped |
| Add group attention merging and group consistency loss (ablation) | DAFL with SSCA on PA100k · PA100k | mA: 82.21 -> 82.65 | Helped |
| Add semantic triplet loss (ablation) | DAFL with SSCA on PA100k · PA100k | mA: 82.21 -> 82.41 | Helped |
| Use group-shared spatial memory instead of separate memory for every attribute (ablation) | DAFL with SSCA and triplet loss on PA100k · PA100k | mA: 83.18 -> 83.54 | Helped |
| Use 2 rather than 1 cascaded SSCA module (ablation) | DAFL on PA100k · PA100k | mA: 82.82 -> 83.31 | Helped |
| Use 3 rather than 2 cascaded SSCA modules (ablation) | DAFL on PA100k · PA100k | mA: 83.31 -> 83.54 | Helped |
| Increase the SSCA cascade from 3 to 4 modules (ablation) | DAFL on PA100k · PA100k | mA: 83.54 -> 83.01 | Hurt |
| Increase the SSCA cascade from 3 to 5 modules (ablation) | DAFL on PA100k · PA100k | mA: 83.54 -> 83.13 | Hurt |
| Increase the SSCA cascade from 3 to 6 modules (ablation) | DAFL on PA100k · PA100k | mA: 83.54 -> 82.73 | Hurt |
| Use DAFL rather than the strongest reported baseline | PETA · PETA | mA: 86.97 -> 87.07 | Helped |
| Use DAFL rather than the strongest reported baseline | PETA · PETA | F1: 87.91 -> 86.40 | Hurt |
| Use DAFL rather than the strongest reported baseline | RAPv1 · RAPv1 | mA: 83.69 -> 83.72 | Helped |
Closest papers by meaning
- Pedestrian Attribute Recognition: A New Benchmark Dataset and a Large Language Model Augmented Framework (AAAI 2025 · similarity 0.89)
- Selective and Orthogonal Feature Activation for Pedestrian Attribute Recognition (AAAI 2024 · similarity 0.89)
- Pedestrian Attribute Recognition as Label-balanced Multi-label Learning (ICML 2024 · similarity 0.88)
- Enhanced Visual-Semantic Interaction with Tailored Prompts for Pedestrian Attribute Recognition (CVPR 2025 · similarity 0.85)
- Attribute-Guided Pedestrian Retrieval: Bridging Person Re-ID with Internal Attribute Variability (CVPR 2024 · similarity 0.85)
- Multispectral Pedestrian Detection with Sparsely Annotated Label (AAAI 2025 · similarity 0.82)
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.