Research map / Vision-language and multimodal / CLIP and zero-shot
Hybrid-Attention Based Decoupled Metric Learning for Zero-Shot Image Retrieval
Check 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 |
|---|---|---|---|
| Use hybrid attention with three object-attention learners and three channel-attention learners | DeML · CUB-200 | R@1: 52.9 -> 65.4 | Helped |
| Use hybrid attention with three object-attention learners and three channel-attention learners | DeML · CARS-196 | R@1: 68.4 -> 86.3 | Helped |
| Add activation decay (ablation) | U512 · CUB | R@1: 52.9 -> 56.2 | Helped |
| Add activation decay (ablation) | U512 · CARS | R@1: 68.4 -> 77.6 | Helped |
| Add a single attention module to the unified metric (ablation) | U512+$L_{act}$ · CUB | R@1: 56.2% vs. 56.1% | No clear effect |
| Increase object-attention learners from 1 to 3 while keeping J=1 (ablation) | DeML · CUB | R@1: 56.1%->60.9%->64.9% | Helped |
| Increase channel-attention sub-learners from 1 to 8 while keeping I=1 (ablation) | DeML · CUB | R@1: 56.1% to 59.0% | Helped |
| Increase channel-attention sub-learners from 1 to 8 while keeping I=1 (ablation) | DeML · CARS | R@1: 77.9% to 82.5% | Helped |
| Remove the adversarial diversity loss (ablation) | DeML(I=1,J=8) · CUB | R@1: 59.0 -> 52.0 | Hurt |
| Remove the adversarial diversity loss (ablation) | DeML(I=1,J=8) · CARS | R@1: 82.5 -> 68.2 | Hurt |
| Remove the adversarial diversity loss (ablation) | DeML(I=1,J=8) · CUB | mean cosine similarity between CA learners: -0.08 ${\pm0.01}$ -> 0.92 {${\pm0.02}$} | Hurt |
Closest papers by meaning
- Energy Confused Adversarial Metric Learning for Zero-Shot Image Retrieval and Clustering (AAAI 2019 · similarity 0.89)
- Dual-View Ranking with Hardness Assessment for Zero-Shot Learning (AAAI 2019 · similarity 0.87)
- Zero-Shot Learning via Joint Latent Similarity Embedding (CVPR 2016 · similarity 0.86)
- Knowledge-Enhanced Dual-stream Zero-shot Composed Image Retrieval (CVPR 2024 · similarity 0.85)
- TPR: Topology-Preserving Reservoirs for Generalized Zero-Shot Learning (NeurIPS 2024 · similarity 0.85)
- Context-I2W: Mapping Images to Context-Dependent Words for Accurate Zero-Shot Composed Image Retrieval (AAAI 2024 · similarity 0.85)
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.