AI Paper Atlas

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

Visual Grounding via Accumulated Attention

Chaorui Deng, Qi Wu, Qingyao Wu, Fuyuan Hu, Fan Lyu, Mingkui Tan

CVPR 2018 · Vision-language and multimodal › Human and embodied modeling · approach: Visual grounding transformers · 215 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
Use A-ATT instead of the strongest listed baselineVisual grounding model · ReferCOCO TestAaccuracy: 78.45 -> 81.67Improved
Use A-ATT instead of the strongest listed baselineVisual grounding model · ReferCOCO TestBaccuracy: 80.10 vs. 80.01No clear effect
Use A-ATT instead of the strongest listed baselineVisual grounding model · ReferCOCO+ TestAaccuracy: 63.34 -> 68.76Improved
Use A-ATT instead of the strongest listed baselineVisual grounding model · ReferCOCOg Valaccuracy: 72.75 -> 73.18Improved
Use A-ATT with word2vector embeddingsGuessWhat?! grounding model · GuessWhat?! Testerror: 36.5% -> 34.2%Improved
Add a second A-ATT round (ablation)A-ATT visual grounding model · ReferCOCO TestAaccuracy: 79.67 -> 81.37Improved
Add a third A-ATT round (ablation)A-ATT visual grounding model · ReferCOCO TestAaccuracy: 81.37 -> 81.67Improved
Add a fourth A-ATT round (ablation)A-ATT visual grounding model · GuessWhat?! Testerror: 35.8% -> 36.3%Hurt
Add query attention (ablation)A-ATT visual grounding model · ReferCOCO TestAaccuracy: 79.14 -> 81.37Improved
Add image attention (ablation)A-ATT visual grounding model · ReferCOCO TestAaccuracy: 79.82 -> 81.37Improved
Add object attention (ablation)A-ATT visual grounding model · ReferCOCO TestAaccuracy: 80.08 -> 81.37Improved
Add interactions among query, image, and object attention (ablation)A-ATT visual grounding model · ReferCOCO TestAaccuracy: 78.09 -> 81.37Improved

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