Research map / Label-efficient and robust learning / Recognition and long-tail
Learning a Discriminative Filter Bank Within a CNN for Fine-Grained 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 |
|---|---|---|---|
| add a second-scale DFL module (ablation) | DFL-CNN with VGG-16 · CUB-200-2011 | accuracy (%): 85.8 -> 86.7 | Helped |
| add a second-scale DFL module (ablation) | DFL-CNN with VGG-16 · Stanford Cars | accuracy (%): 93.3 -> 93.8 | Helped |
| add a second-scale DFL module (ablation) | DFL-CNN with VGG-16 · FGVC-Aircraft | accuracy (%): 91.1 -> 92.0 | Helped |
| combine global and patch streams instead of using either alone (ablation) | DFL-CNN at test time · CUB-200-2011 | accuracy (%): 80.3 / 82.0 -> 84.9 | Helped |
| add the side branch to the combined global and patch streams (ablation) | DFL-CNN at test time · CUB-200-2011 | accuracy (%): 84.9 -> 85.8 | Helped |
| replace GMP with GAP (ablation) | DFL-CNN patch stream · CUB-200-2011 | accuracy (%): 85.8 -> 80.4 | Hurt |
| add non-random layer initialization (ablation) | DFL-CNN without filter supervision · CUB-200-2011 | accuracy (%): 82.2 -> 84.4 | Helped |
| add filter supervision to non-random layer initialization (ablation) | DFL-CNN · CUB-200-2011 | accuracy (%): 84.4 -> 85.8 | Helped |
| use bounding-box annotations (ablation) | DFL-CNN · CUB-200-2011 | accuracy (%): 85.8 -> 85.7 | No clear effect |
Closest papers by meaning
- Fine-Grained Recognition: Accounting for Subtle Differences between Similar Classes (AAAI 2020 · similarity 0.90)
- Part-Stacked CNN for Fine-Grained Visual Categorization (CVPR 2016 · similarity 0.89)
- Discrimination-Aware Mechanism for Fine-Grained Representation Learning (CVPR 2021 · similarity 0.88)
- PICNN: A Pathway towards Interpretable Convolutional Neural Networks (AAAI 2024 · similarity 0.86)
- Twofold Debiasing Enhances Fine-Grained Learning with Coarse Labels (AAAI 2025 · similarity 0.86)
- AsCAN: Asymmetric Convolution-Attention Networks for Efficient Recognition and Generation (NeurIPS 2024 · 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.