Research map / Vision-language and multimodal / Human and embodied modeling
Similarity Reasoning and Filtration for Image-Text Matching
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 |
|---|---|---|---|
| combine SGR and SAF with independent training (ablation) | SGRAF · MSCOCO 1K test, sentence retrieval | R@1: 77.8 -> 79.6 | Helped |
| combine SGR and SAF with independent training (ablation) | SGRAF · Flickr30K test, sentence retrieval | R@1: 75.1 -> 77.8 | Helped |
| add graph reasoning to global alignment (ablation) | SGR · Flickr30K, sentence retrieval | R@1: 62.4 -> 75.2 | Helped |
| add graph reasoning to global alignment (ablation) | SGR · Flickr30K, image retrieval | R@1: 46.0 -> 56.2 | Helped |
| add global similarity to local alignment reasoning (ablation) | SGR · Flickr30K | R@1 | Helped |
| use textual-to-visual rather than visual-to-textual local attention (ablation) | SGR · Flickr30K, sentence retrieval | R@1: 66.1 -> 73.6 | Helped |
| use SAF instead of average aggregation for local alignments (ablation) | similarity aggregation · Flickr30K, sentence retrieval | R@1: 67.2 -> 68.2 | Helped |
| use vector-based instead of scalar cosine similarity (ablation) | similarity representation · Flickr30K, sentence retrieval | R@1: 66.7 -> 67.2 | Helped |
| set graph reasoning to 3 steps (ablation) | SGR · Flickr30K, sentence retrieval | R@1: 74.2, 75.3, 75.2, 76.2 | Mixed |
| use graph dimension 256 rather than 128 (ablation) | SGR · Flickr30K, sentence retrieval | R@1: 74.5 -> 75.2 | Helped |
| use SAF instead of SGR alone (ablation) | SGRAF modules · Flickr30K, sentence retrieval | R@1: 75.2 -> 73.7 | Mixed |
Closest papers by meaning
- Composing Object Relations and Attributes for Image-Text Matching (CVPR 2024 · similarity 0.87)
- GOAL: Global-local Object Alignment Learning (CVPR 2025 · similarity 0.86)
- Visual-Text Cross Alignment: Refining the Similarity Score in Vision-Language Models (ICML 2024 · similarity 0.86)
- Identification of Necessary Semantic Undertakers in the Causal View for Image-Text Matching (AAAI 2024 · similarity 0.86)
- Text Embedding is Not All You Need: Attention Control for Text-to-Image Semantic Alignment with Text Self-Attention Maps (CVPR 2025 · similarity 0.85)
- Text as Partial Constraint: Core–Residual Alignment for Robust Vision–Language Learning (NeurIPS 2026 · 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.