AI Paper Atlas

Research map / Graph learning / Knowledge and temporal graphs

Unconditional Scene Graph Generation

Sarthak Garg, Helisa Dhamo, Azade Farshad, Sabrina Musatian, Nassir Navab, Federico Tombari

ICCV 2021 · Graph learning › Knowledge and temporal graphs · approach: Scene graph generation · 21 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 node information and E_from/E_to conditioning to edge generation (SceneGraphGen vs GraphRNN, BFS) (ablation)SceneGraphGen · Visual Genome graphsMMD graph (x10^3): 1.3 -> 1.82Mixed
use SceneGraphGen instead of GraphRNN with random orderingGraphRNN · Visual Genome graphsMMD graph (x10^3): 1.2 -> 0.11Improved
use SceneGraphGen instead of GraphRNN with random orderingGraphRNN · Visual Genome graphsMMD node (x10^3): 0.39 -> 0.37No clear effect
use random node ordering instead of BFS ordering (ablation)SceneGraphGen · Visual Genome graphsMMD graph (x10^3): 1.82 -> 0.11Improved
use random node ordering instead of hierarchical ordering (ablation)SceneGraphGen · Visual Genome graphsMMD graph (x10^3): 0.63 -> 0.11Improved
use random node ordering instead of BFS ordering (ablation)SceneGraphGen · Visual Genome graphsMMD node (x10^3): 2.05 -> 0.37Improved
use random node ordering instead of hierarchical ordering (ablation)SceneGraphGen · Visual Genome imagesFID: 72.2 -> 71.2Improved
use random node ordering instead of hierarchical ordering (ablation)SceneGraphGen · Visual Genome imagesIS: 5.26 -> 4.95Hurt
use SceneGraphGen graphs instead of GraphRNN graphs to drive sg2imGraphRNN · Visual Genome imagesFID: 74.5 -> 71.2Improved
use scene graphs plus sg2im instead of unconditional StyleGAN2StyleGAN2 · Visual Genome imagesFID: 66.3 -> 71.2Hurt
use scene graphs plus sg2im instead of unconditional StyleGAN2StyleGAN2 · Visual Genome imagesdetected object categories: 40 -> 50Improved
use scene graphs plus sg2im instead of unconditional StyleGAN2StyleGAN2 · Visual Genome imagesobject occurrence error vs ground truth: 1.4 -> 1.2Improved

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