AI Paper Atlas

Research map / Graph learning / Knowledge and temporal graphs

Say As You Wish: Fine-Grained Control of Image Caption Generation With Abstract Scene Graphs

Shizhe Chen, Qin Jin, Peng Wang, Qi Wu

CVPR 2020 · Graph learning › Knowledge and temporal graphs · approach: Scene graph generation · 263 citations (OpenAlex)

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 ASG2Caption instead of C-BUTDControllable image captioning · VisualGenomeCIDEr: 159.5 -> 202.4Improved
Use ASG2Caption instead of C-BUTDControllable image captioning · MSCOCOCIDEr: 143.8 -> 204.2Improved
Use ASG2Caption instead of C-BUTDControllable image captioning · VisualGenomeG_a: 0.7 -> 0.3Improved
Use ASG2Caption instead of C-BUTDControllable image captioning · MSCOCOG_a: 1.0 -> 0.3Improved
Generate captions using sampled ASGs rather than the comparison baselineDiverse caption generation · VisualGenomeSelfCIDEr: 0.47 -> 0.75Improved
Generate captions using sampled ASGs rather than BUTD beam searchDiverse caption generation · MSCOCOSelfCIDEr: 0.58 -> 0.76Improved
Add role-aware node embeddings (ablation)C-BUTD baseline · VisualGenomeCIDEr: 146.3 -> 176.9Improved
Add role-aware node embeddings (ablation)C-BUTD baseline · MSCOCOCIDEr: 135.7 -> 166.9Improved
Add MR-GCN contextual graph encoding (ablation)Role-aware encoder with graph content attention · VisualGenomeCIDEr: 176.9 -> 191.7Improved
Add MR-GCN contextual graph encoding (ablation)Role-aware encoder with graph content attention · MSCOCOCIDEr: 166.9 -> 190.5Improved
Add graph flow attention (ablation)Model with role embeddings, MR-GCN, and graph content attention · VisualGenomeCIDEr: 191.7 -> 193.1Improved
Add graph flow attention (ablation)Model with role embeddings, MR-GCN, and graph content attention · MSCOCOCIDEr: 190.5 -> 196.2Improved

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