AI Paper Atlas

Research map / Vision-language and multimodal / Human and embodied modeling

Negative-Aware Attention Framework for Image-Text Matching

Kun Zhang, Zhendong Mao, Quan Wang, Yongdong Zhang

CVPR 2022 · Vision-language and multimodal › Human and embodied modeling · approach: Language-grounded scene text and structure · 204 citations (OpenAlex)

Check 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
use negative-aware attention with positive and negative branchesNAAF · Flickr30K testImage-to-text R@1: 77.8 -> 81.9Improved
use negative-aware attention with positive and negative branchesNAAF · MS-COCO 1K testImage-to-text R@1: 79.6 -> 80.5Improved
use negative-aware attention with positive and negative branchesNAAF · MS-COCO 5K testImage-to-text R@1: 53.7 -> 58.9Improved
add the negative attention branch to capture mismatched-fragment effects (ablation)NAAF · Flickr30KImage-to-text R@1: 66.2 -> 81.9Improved
add the designed positive and negative attention masks (ablation)NAAF · Flickr30KImage-to-text R@1: 74.2 -> 81.9Improved
add the image-to-text matching direction (ablation)NAAF · Flickr30KImage-to-text R@1: 81.9 -> 78.7Hurt
use GloVe word embeddings (ablation)NAAF · Flickr30KImage-to-text R@1: 75.9 -> 81.9Improved
add intra-modal word relationships (ablation)NAAF · Flickr30KImage-to-text R@1: 79.3 -> 81.9Improved
set initial mismatch penalty α to 3.0 instead of 1.0 (ablation)NAAF · Flickr30KImage-to-text R@1: 78.0 -> 78.9Improved
adjust α to α* during training (ablation)NAAF · Flickr30KImage-to-text R@1: 78.7 -> 79.6Improved
increase batch size from 32 to 128 (ablation)NAAF · Flickr30KImage-to-text R@1: 76.0 -> 79.6Improved
update negative samples during training (ablation)NAAF · Flickr30KImage-to-text R@1: 76.3 -> 79.6Improved

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