Research map / Label-efficient and robust learning / Recognition and long-tail
Channel Interaction Networks for Fine-Grained Image Categorization
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 SCI to ResNet-50 (ablation) | ResNet-50 · CUB-200-2011 | top-1 accuracy: 84.9% -> 87.1%; 2.2% | Helped |
| replace SCI with SE (ablation) | ResNet-50 + SCI · CUB-200-2011 | top-1 accuracy: 87.1% vs. 85.7% | Hurt |
| use positive SCI weights without the negative sign (ablation) | ResNet-50 + SCI · CUB-200-2011 | top-1 accuracy: 87.1% vs. 86.1% | Hurt |
| add the CCI module for pair-conditioned channel interaction (ablation) | ResNet-50 + SCI · CUB-200-2011 | top-1 accuracy: 87.1% -> 87.5%; 0.4% | Helped |
| apply contrastive loss to SCI features without pair-conditioned CCI (ablation) | ResNet-50 + SCI · CUB-200-2011 | top-1 accuracy: 87.1% vs. 87.2% | No clear effect |
| use triplet loss instead of contrastive loss in CCI (ablation) | CIN · fine-grained image categorization | top-1 accuracy | No clear effect |
| use addition, multiplication, or concatenation instead of subtraction for CCI weights (ablation) | CIN · fine-grained image categorization | top-1 accuracy | Hurt |
| use all O(n²) image pairs or hard-negative mining (ablation) | CIN training · fine-grained image categorization | top-1 accuracy | Hurt |
| add CIN to NTS | NTS · CUB-200-2011 | top-1 accuracy: 87.5% -> 88.3% | Helped |
| add CIN to NTS | NTS · FGVC Aircraft | top-1 accuracy: 91.4% -> 93.3% | Helped |
| add CIN to NTS | NTS · Stanford Cars | top-1 accuracy: 93.9% -> 94.4% | Helped |
Closest papers by meaning
- Learning Attentive Pairwise Interaction for Fine-Grained Classification (AAAI 2020 · similarity 0.88)
- Fine-Grained Recognition: Accounting for Subtle Differences between Similar Classes (AAAI 2020 · similarity 0.86)
- Learning Channel-Wise Interactions for Binary Convolutional Neural Networks (CVPR 2019 · similarity 0.86)
- Fine-Grained Image Classification via Combining Vision and Language (CVPR 2017 · similarity 0.85)
- $\mathbb{X}$-Sample Contrastive Loss: Improving Contrastive Learning with Sample Similarity Graphs (ICLR 2025 · similarity 0.84)
- Cross-Layer and Cross-Sample Feature Optimization Network for Few-Shot Fine-Grained Image Classification (AAAI 2024 · similarity 0.84)
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.