AI Paper Atlas

Research map / Vision-language and multimodal / Cross-modal learning

Saliency-Guided Attention Network for Image-Sentence Matching

Zhong Ji, Haoran Wang, Jungong Han, Yanwei Pang

ICCV 2019 · Vision-language and multimodal › Cross-modal learning · approach: Contrastive vision-language alignment · 121 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (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 changedWhereMeasuredEffect
replace global visual feature with SVA saliency feature (ablation)SAN (no STA) · MSCOCO 1K testsentence retrieval R@1: 63.4 -> 65.4Improved
replace global visual feature with SVA saliency feature (ablation)SAN (no STA) · MSCOCO 1K testimage retrieval R@1: 50.5 -> 52.9Improved
fuse global visual feature with SVA feature (ablation)SAN (no STA) · MSCOCO 1K testsentence retrieval R@1: 65.4 -> 66.2Improved
fuse global visual feature with SVA feature (ablation)SAN (no STA) · MSCOCO 1K testimage retrieval R@1: 52.9 -> 53.7Improved
use self-attention textual feature (IT) on top of fused visual feature (ablation)SAN (FV + GT) · MSCOCO 1K testimage retrieval R@1: 53.7 -> 56.6Improved
merge global and self-attention textual features (FT G-I) (ablation)SAN (FV + GT) · MSCOCO 1K testsentence retrieval R@1: 66.2 -> 66.8No clear effect
add STA module (FT G-S) to fused visual baseline (ablation)SAN (FV + GT) · MSCOCO 1K testsentence retrieval R@1: 66.2 -> 85.4Improved
add STA module (FT G-S) to fused visual baseline (ablation)SAN (FV + GT) · MSCOCO 1K testimage retrieval R@1: 53.7 -> 69.1Improved
use SVA feature instead of global feature as STA visual guidance (ablation)SAN (STA) · MSCOCO 1K testsentence retrieval R@1: 74.5 -> 82.1Improved
use SVA feature instead of global feature as STA visual guidance (ablation)SAN (STA) · MSCOCO 1K testimage retrieval R@1: 57.8 -> 67.3Improved
fuse global and SVA visual features as STA guidance (ablation)SAN (STA) · MSCOCO 1K testsentence retrieval R@1: 82.1 -> 85.4Improved
fuse global and SVA visual features as STA guidance (ablation)SAN (STA) · MSCOCO 1K testimage retrieval R@1: 67.3 -> 69.1Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy