Research map / Label-efficient and robust learning / Knowledge distillation
GLAMD: Global and Local Attention Mask Distillation for Object Detectors
Younho Jang, Wheemyung Shin, Jinbeom Kim, Simon Woo, Sung-Ho Bae
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 GLAM mask to feature distillation (ablation) | RetinaNet R50 student · COCO2017 val (RetinaNet R50) | AP: 37.1 -> 39.5 | Improved |
| add classification head distillation (ablation) | GLAMD RetinaNet · COCO2017 val (RetinaNet R50) | AP: 39.5 -> 39.7 | Improved |
| add localization head distillation (no cls head) (ablation) | GLAMD RetinaNet · COCO2017 val (RetinaNet R50) | AP: 39.5 -> 39.6 | Improved |
| add both cls and loc head distillation (ablation) | GLAMD RetinaNet · COCO2017 val (RetinaNet R50) | AP: 39.5 -> 40.0 | Improved |
| add local attention mask alone (ablation) | attention-mask KD · COCO2017 val (RetinaNet R50) | AP: 38.7 -> 39.8 | Improved |
| add global attention mask alone (ablation) | attention-mask KD · COCO2017 val (RetinaNet R50) | AP: 38.7 -> 38.9 | Improved |
| combine global with local attention mask (ablation) | local-only GLAMD · COCO2017 val (RetinaNet R50) | AP: 39.8 -> 40.0 | Improved |
| reduce patch size from 9 to 7 (ablation) | GLAMD · COCO2017 val (RetinaNet R50) | AP: 39.8 -> 40.0 | Improved |
| reduce patch size from 7 to 3 (ablation) | GLAMD · COCO2017 val (RetinaNet R50) | AP: 40.0 -> 39.7 | Hurt |
| use IoU loss instead of L1 for loc head distillation (ablation) | GLAMD · COCO2017 val (RetinaNet R50) | AP: 39.9 -> 40.0 | Improved |
| use IoU loss instead of MSE or smooth-L1 (ablation) | GLAMD · COCO2017 val (RetinaNet R50) | AP: 39.8 -> 40.0 | Improved |
| GLAMD vs global-attention KD of Zhang et al. | Faster R-CNN R50 · COCO2017 val | AP: 40.1 -> 40.8 | Improved |
Closest papers by meaning
- Improve Object Detection with Feature-based Knowledge Distillation: Towards Accurate and Efficient Detectors (ICLR 2021 · similarity 0.92)
- When Object Detection Meets Knowledge Distillation: A Survey (TPAMI 2023 · similarity 0.91)
- ACAM-KD: Adaptive and Cooperative Attention Masking for Knowledge Distillation (ICCV 2025 · similarity 0.91)
- DCSF-KD: Dynamic Channel-wise Spatial Feature Knowledge Distillation for Object Detection (AAAI 2025 · similarity 0.91)
- DFD: Distilling the Feature Disparity Differently for Detectors (ICML 2024 · similarity 0.90)
- CrossKD: Cross-Head Knowledge Distillation for Object Detection (CVPR 2024 · similarity 0.89)
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.
