Research map / Label-efficient and robust learning / Recognition and long-tail
Residual Attention: A Simple but Effective Method for Multi-Label Recognition
Ke Zhu, Jianxin Wu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 class-specific spatial pooling to average pooling (ablation) | CSRA ResNet-cut · MS-COCO | mAP (%): 82.1 -> 85.3 | Improved |
| use spatial pooling alone versus combining with average pooling (ablation) | CSRA ResNet-cut · MS-COCO | mAP (%): 84.2 -> 85.3 | Improved |
| add the residual average-pooling term to spatial pooling (ablation) | CSRA ResNet-cut · MS-COCO | mAP (%): 82.1 -> 84.2 | Improved |
| increase attention heads from 1 to 8 (ablation) | CSRA VIT-L16 · MS-COCO | mAP (%): 85.8 -> 86.5 | Improved |
| increase attention heads from 1 to 6 (ablation) | CSRA ResNet-cut · MS-COCO | mAP (%): 85.3 -> 85.6 | Improved |
| add classifier normalization (H=2) (ablation) | CSRA ResNet-101 · MS-COCO | mAP (%): 83.1 -> 83.2 | No clear effect |
| add classifier normalization (H=4) (ablation) | CSRA ResNet-101 · MS-COCO | mAP (%): 83.4 -> 83.3 | No clear effect |
| add CSRA to VIT-L16 | VIT-L16 · MS-COCO | mAP (%): 80.4 -> 86.5 | Improved |
| add CSRA to ResNet-101 | ResNet-101 · MS-COCO | mAP (%): 79.4 -> 83.5 | Improved |
| add CSRA to VIT-B16 | VIT-B16 · WIDER-Attribute | mAP (%): 86.3 -> 89.0 | Improved |
| add RandAugment to CSRA | VIT-L16 + CSRA · MS-COCO | mAP (%): 86.5 -> 86.9 | Improved |
| add training-free test-time global max pooling to average pooling | MobileNet · MS-COCO | mAP (%): 71.8 -> 73.0 | Improved |
Closest papers by meaning
- Multi-Label Image Recognition by Recurrently Discovering Attentional Regions (ICCV 2017 · similarity 0.89)
- DualCoOp++: Fast and Effective Adaptation to Multi-Label Recognition With Limited Annotations (TPAMI 2023 · similarity 0.88)
- [CLS] is Not Enough: Multi-Label Recognition via Patch-Level Inference and Adaptive Aggregation (ICML 2026 · similarity 0.88)
- A Shared Multi-Attention Framework for Multi-Label Zero-Shot Learning (CVPR 2020 · similarity 0.87)
- In Pursuit of Causal Label Correlations for Multi-label Image Recognition (NeurIPS 2024 · similarity 0.86)
- Correlative and Discriminative Label Grouping for Multi-Label Visual Prompt Tuning (CVPR 2025 · 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.
