AI Paper Atlas

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

Adversarial Representation Learning for Text-to-Image Matching

Nikolaos Sarafianos, Xiang Xu, Ioannis A. Kakadiaris

ICCV 2019 · Vision-language and multimodal › Cross-modal learning · approach: Probabilistic embeddings for cross-modal retrieval · 259 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 cross-modal projection matching loss to identification loss (ablation)TIMAM · CUHK-PEDES text-to-imageRank-1: 40.1 -> 49.8Improved
add identification loss to projection matching loss (ablation)TIMAM · CUHK-PEDES text-to-imageRank-1: 44.9 -> 49.8Improved
add BERT word embeddings (ablation)TIMAM (no ARL) · CUHK-PEDES text-to-imageRank-1: 49.8 -> 52.9Improved
add adversarial representation learning (ablation)TIMAM (no BERT) · CUHK-PEDES text-to-imageRank-1: 49.8 -> 51.3Improved
add adversarial representation learning (ablation)TIMAM (with BERT) · CUHK-PEDES text-to-imageRank-1: 52.9 -> 54.5Improved
add adversarial representation learning (ablation)TIMAM · Flickr30K image-to-textRank-1: 51.2 -> 53.1Improved
add adversarial representation learning (ablation)TIMAM · Flickr30K text-to-imageRank-1: 41.0 -> 42.6Improved
replace FC word embedding with BERT (ablation)TIMAM (ResNet-101) · Flickr30K text-to-imageRank-1: 35.8 -> 40.6Improved
replace FC word embedding with BERT (ablation)TIMAM (ResNet-152) · Flickr30K image-to-textRank-1: 50.1 -> 53.1Improved
deepen image backbone from ResNet-101 to ResNet-152 (ablation)TIMAM (BERT) · Flickr30K text-to-imageRank-1: 40.6 -> 42.6Improved
add conditional text augmentation (ablation)TIMAM · CUHK-PEDES text-to-imageRank-1Hurt
add text-to-image reconstruction objective (ablation)TIMAM · CUHK-PEDES and Flickr30KRank-1No clear effect

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