Research map / Label-efficient and robust learning / Recognition and long-tail
Fine-Grained Recognition: Accounting for Subtle Differences between Similar Classes
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 |
|---|---|---|---|
| add the diversification block to ResNet-50 (ablation) | ResNet-50 · CUB-200-2011 | accuracy: 85.5 -> 86.3 | Helped |
| replace gradient-boosting loss with Center loss (ablation) | ResNet-50+DB · CUB-200-2011 | accuracy: 86.3 -> 86.4 | Hurt |
| replace gradient-boosting loss with LGM loss (ablation) | ResNet-50+DB · CUB-200-2011 | accuracy: 86.3 -> 86.5 | Hurt |
| replace gradient-boosting loss with MaxEnt (ablation) | ResNet-50+DB · CUB-200-2011 | accuracy: 86.3 -> 86.2 | Hurt |
| use gradient-boosting loss instead of the baseline loss (ablation) | ResNet-50+DB · CUB-200-2011 | accuracy: 86.3% to 87.7% | Helped |
| set suppressing factor α to 0 (ablation) | Diversification block · CUB-200-2011 | accuracy: 86.2 | No clear effect |
| set suppressing factor α to 0.1 (ablation) | Diversification block · CUB-200-2011 | accuracy: 86.3 | Helped |
| set suppressing factor α to 0.2 (ablation) | Diversification block · CUB-200-2011 | accuracy: 85.8 | Hurt |
| disable suppression by setting α to 1.0 (ablation) | Diversification block · CUB-200-2011 | accuracy: 86.3 -> 85.5 | Hurt |
| use multi-scale instead of single-scale inference | Ours · CUB-200-2011 | accuracy: 87.7 -> 88.6 | Helped |
| use gradient-boosting loss rather than cross entropy | ResNet-50 · ImageNet | error rate | Helped |
Closest papers by meaning
- Discrimination-Aware Mechanism for Fine-Grained Representation Learning (CVPR 2021 · similarity 0.90)
- Fine-grained Classes and How to Find Them (ICML 2024 · similarity 0.88)
- Twofold Debiasing Enhances Fine-Grained Learning with Coarse Labels (AAAI 2025 · similarity 0.88)
- From Coarse to Fine-Grained Open-Set Recognition (CVPR 2024 · similarity 0.87)
- Finer-CAM: Spotting the Difference Reveals Finer Details for Visual Explanation (CVPR 2025 · similarity 0.87)
- Bidirectional Logits Tree: Pursuing Granularity Reconcilement in Fine-Grained Classification (AAAI 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.