AI Paper Atlas

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

Learning Two-Branch Neural Networks for Image-Text Matching Tasks

Liwei Wang, Yin Li, Jing Li Huang, Svetlana Lazebnik

TPAMI 2018 · Vision-language and multimodal › Human and embodied modeling · approach: Language-grounded scene text and structure · 546 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
add nonlinear (two-layer) branches instead of linear (ablation)embedding network · Flickr30K Entities phrase localizationR@1: 39.60 -> 43.15Improved
use bi-directional instead of single-directional loss (ablation)embedding network · Flickr30K Entities phrase localizationR@1: 43.15 -> 44.53Improved
add positive region augmentation (IoU > 0.7) (ablation)embedding network · Flickr30K Entities phrase localizationR@1: 44.53 -> 49.09Improved
add neighborhood sampling (ablation)embedding network · Flickr30K Entities phrase localizationR@1: 49.09 -> 51.03Improved
add neighborhood constraints on top of sampling (ablation)embedding network · Flickr30K Entities phrase localizationR@1: 51.03 -> 50.69No clear effect
add positive region augmentation (ablation)similarity network · Flickr30K Entities phrase localizationR@1: 45.19 -> 51.05Improved
add nonlinear branches instead of linear (ablation)similarity network · Flickr30K Entities phrase localizationR@1: 43.19 -> 45.19Improved
use three FC layers after element-wise product instead of one (ablation)similarity network · Flickr30K Entities phrase localizationR@1: 36.61 -> 51.05Improved
use bi-directional instead of single-directional loss (ablation)embedding network · Flickr30K sentence-to-imageR@1: 22.1 -> 30.7Improved
add neighborhood sampling (ablation)embedding network · MSCOCO 1000 image-to-sentenceR@1: 53.0 -> 54.2Improved
add neighborhood constraints on top of sampling (ablation)embedding network · MSCOCO 1000 image-to-sentenceR@1: 54.2 -> 54.9Improved
add neighborhood constraints on top of sampling (ablation)embedding network · Flickr30K sentence-to-sentenceR@1: 60.5 -> 63.8Improved

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