AI Paper Atlas

Research map / Detection and segmentation / Open-set and few-shot detection

Attributes as Operators: Factorizing Unseen Attribute-Object Compositions

Tushar Nagarajan, Kristen Grauman

ECCV 2018 · Detection and segmentation › Open-set and few-shot detection · approach: Object-centric learning theory and video · 183 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 auxiliary classifier loss (ablation)base attribute-operator model · MIT-Statesharmonic mean accuracy (%): 3.7 -> 9.9Improved
add auxiliary classifier loss (ablation)base attribute-operator model · UT-Zapposharmonic mean accuracy (%): 20.4 -> 29.5Improved
add inverse consistency regularizer (ablation)base model · MIT-Statesharmonic mean accuracy (%): 3.7 -> 4.5Improved
add inverse consistency regularizer (ablation)base model · UT-Zapposharmonic mean accuracy (%): 20.4 -> 21.7Improved
add inverse consistency regularizer (ablation)model with auxiliary loss · UT-Zapposharmonic mean accuracy (%): 29.5 -> 29.2No clear effect
add commutative regularizer (ablation)model with auxiliary loss · MIT-Statesharmonic mean accuracy (%): 9.9 -> 11.1Improved
add commutative regularizer (ablation)model with auxiliary loss · UT-Zapposharmonic mean accuracy (%): 29.5 -> 33.4Improved
add antonym consistency regularizer (ablation)model with auxiliary loss · MIT-Statesharmonic mean accuracy (%): 9.9 -> 8.8Hurt
add commutative regularizer (ablation)model with auxiliary and inverse losses · UT-Zapposharmonic mean accuracy (%): 29.2 -> 27.5Hurt
add oracle object labels at test timeAttributes as operators · MIT-Statesopen world accuracy (%): 11.4 -> 49.3Improved
add auxiliary loss and trainable inputs to baseline (ablation)LabelEmbed+ · MIT-Statesharmonic mean accuracy (%): 8.2 -> 7.7Mixed
increase LabelEmbed+ layers from 1 to 3 (ablation)LabelEmbed+ · MIT-Statesharmonic mean accuracy (%): 8.3 -> 7.7No 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