Research map / Image generation / Autoregressive and latent generation
Composed Query Image Retrieval Using Locally Bounded Features
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 |
|---|---|---|---|
| represent images as local regions and correlate text words with image regions | composed query image retrieval · Fashion200K | Recall@1: 14.1±0.6 -> 17.78±0.5 | Helped |
| represent images as local regions and correlate text words with image regions | composed query image retrieval · MIT States | Recall@1: 12.3±0.5 -> 14.72±0.6 | Helped |
| represent images as local regions and correlate text words with image regions | composed query image retrieval · CSS 3D→3D | Recall@1: 73.7±1.0 -> 79.2±1.2 | Helped |
| represent images as local regions and correlate text words with image regions | composed query image retrieval · CSS 2D→3D | Recall@1: 46.6 -> 55.69±0.9 | Helped |
| add the auxiliary loss LB during training (ablation) | proposed method · CSS 3D | Recall@1: 76.1 -> 79.2 | Helped |
| use 36 rather than 18 region proposals (ablation) | proposed method · Fashion200K | Recall@1: 16.26±0.6 -> 17.78±0.5 | Helped |
| use 36 rather than 18 region proposals (ablation) | proposed method · MIT States | Recall@1: 14.29±0.6 -> 14.72±0.6 | Helped |
| use 36 rather than 18 region proposals (ablation) | proposed method · CSS 3D→3D | Recall@1: 67.26±1.1 -> 79.2±1.2 | Helped |
| use coarse filtering with θ = 100 (ablation) | proposed method at inference · CSS 3D | Recall@1: 79.2 -> 75.92 | Hurt |
| use coarse filtering with θ = 100 (ablation) | proposed method at inference · CSS 3D | ACRR: 75.24% | Helped |
| use coarse filtering with θ = 85 (ablation) | proposed method at inference · CSS 3D | Recall@1: 79.2 -> 70.81 | Hurt |
| use coarse filtering with θ = 85 (ablation) | proposed method at inference · CSS 3D | ACRR: 91.86% | Helped |
Closest papers by meaning
- Instance-Level Composed Image Retrieval (NeurIPS 2025 · similarity 0.88)
- ConText-CIR: Learning from Concepts in Text for Composed Image Retrieval (CVPR 2025 · similarity 0.87)
- Data Roaming and Quality Assessment for Composed Image Retrieval (AAAI 2024 · similarity 0.87)
- Generative Zero-Shot Composed Image Retrieval (CVPR 2025 · similarity 0.86)
- Decomposing Semantic Shifts for Composed Image Retrieval (AAAI 2024 · similarity 0.86)
- Imagine and Seek: Improving Composed Image Retrieval with an Imagined Proxy (CVPR 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.