Research map / Label-efficient and robust learning / Recognition and long-tail
Semantic-Guided Fine-Tuning of Foundation Model for Long-Tailed Visual Recognition
Yufei Peng, Yonggang Zhang, Yiu‐ming Cheung
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 SG-Adapter (ablation) | baseline without SG · CIFAR-100-LT (imbalance 10) | top-1 accuracy: 84.9 -> 86.0 | Improved |
| add SG-Adapter (ablation) | baseline without SG · CIFAR-100-LT (imbalance 50) | top-1 accuracy: 83.1 -> 84.1 | Improved |
| add SG-Adapter (ablation) | baseline without SG · CIFAR-100-LT (imbalance 100) | top-1 accuracy: 81.7 -> 81.9 | Improved |
| add SG-Adapter (ablation) | baseline without SG · ImageNet-LT | top-1 accuracy: 78.3 -> 78.4 | No clear effect |
| add text-average initialization (Init) (ablation) | SG-Adapter · CIFAR-100-LT (imbalance 100) | top-1 accuracy: 81.9 -> 82.5 | Improved |
| add compensation factor (CF) (ablation) | SG + Init · CIFAR-100-LT (imbalance 100) | top-1 accuracy (tail): 77.2 -> 78.3 | Improved |
| add compensation factor (CF) (ablation) | SG + Init · Places-LT | top-1 accuracy (tail): 52.2 -> 53.2 | Improved |
| add feature interchange technique (FIT) (ablation) | SG + Init + CF · CIFAR-100-LT (imbalance 100) | top-1 accuracy (tail): 78.3 -> 80.0 | Improved |
| add feature interchange technique (FIT) (ablation) | SG + Init + CF · ImageNet-LT | top-1 accuracy (head): 81.2 -> 82.1 | Improved |
| add feature interchange technique (FIT) (ablation) | SG + Init + CF · iNaturalist 2018 | top-1 accuracy: 80.9 -> 80.9 | No clear effect |
| add feature interchange technique (FIT) (ablation) | SG + Init + CF · Places-LT | top-1 accuracy (medium): 53.3 -> 52.8 | Hurt |
| add compensation factor (CF) (ablation) | CE · CIFAR-100-LT (imbalance 100) | top-1 accuracy (tail): 11.5 -> 24.7 | Improved |
Closest papers by meaning
- Rethinking Class-Balanced Methods for Long-Tailed Visual Recognition From a Domain Adaptation Perspective (CVPR 2020 · similarity 0.89)
- Deep Long-Tailed Learning: A Survey (TPAMI 2023 · similarity 0.89)
- Long-Tailed Visual Recognition via Self-Heterogeneous Integration With Knowledge Excavation (CVPR 2023 · similarity 0.88)
- Improving Visual Prompt Tuning by Gaussian Neighborhood Minimization for Long-Tailed Visual Recognition (NeurIPS 2024 · similarity 0.88)
- Decoupled Optimisation for Long-Tailed Visual Recognition (AAAI 2024 · similarity 0.87)
- Feature Fusion from Head to Tail for Long-Tailed Visual Recognition (AAAI 2024 · similarity 0.87)
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.
