AI Paper Atlas

Research map / Graph learning / Knowledge and temporal graphs

Learning Visual Commonsense for Robust Scene Graph Generation

Alireza Zareian, Zhecan Wang, Haoxuan You, Shih-Fu Chang

ECCV 2020 · Graph learning › Knowledge and temporal graphs · approach: Scene graph generation · 70 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 local attention heads to a global-only transformer (GLAT) (ablation)scene graph transformer · VG masked reconstructionoverall node accuracy (%): 22.9 -> 34.4Improved
add global heads to a local-only graph attention net (GLAT) (ablation)scene graph transformer · VG masked reconstructionoverall node accuracy (%): 22.3 -> 34.4Improved
learned attention instead of fixed equal attention on neighbors (ablation)graph network · VG masked reconstructionoverall node accuracy (%): 19.7 -> 22.3Improved
GLAT instead of triplet frequency prior for masked predicatescommonsense model · VG masked reconstructionpredicate accuracy (%): 44.4 -> 60.7Improved
stack GLAT on IMPIMP · VG SGG PredClsmR@50: 9.8 -> 11.1Improved
add fusion module to IMP + GLAT (ablation)IMP + GLAT · VG SGG PredClsmR@50: 11.1 -> 12.1Improved
stack GLAT on SNMSNM · VG SGG PredClsmR@50: 13.3 -> 13.6No clear effect
add fusion module to SNM + GLAT (ablation)SNM + GLAT · VG SGG PredClsmR@50: 13.6 -> 14.1Improved
stack GLAT on KERNKERN · VG SGG PredClsmR@50: 17.7 -> 17.6No clear effect
add fusion module to KERN + GLAT (ablation)KERN + GLAT · VG SGG SGClsmR@50: 9.3 -> 9.9Improved
add GLAT and fusion to IMPIMP · VG SGG SGClsmR@100: 6.0 -> 7.0Improved
add GLAT and fusion to IMPIMP · VG SGCls R@100 by triplet frequency (average)R@100 (%): 36.1 -> 37.4Improved

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