MIT Press

Skip to Main Content

Table 2:

CiC-Zero Shot Generalization. Zero-shot generalization to unseen objects on the Chairs-in-Context (CiC) dataset. Results suggest NES can learn words as event classifiers in a general, object-agnostic manner. *SG model from (Achlioptas et al., 2019).

Model	Zero-Shot Classes				All
Model	Lamp	Bed	Table	Sofa	All
Major.	0.333	0.333	0.333	0.333	0.333
*SG	0.501	0.564	0.637	0.536	0.560
PoE	0.422	0.466	0.587	0.483	0.490
NMN	0.462	0.492	0.572	0.532	0.515
MAC	0.533	0.531	0.632	0.551	0.567
NES
w/VGG16	0.544	0.578	0.693	0.588	0.601
w/Res101	0.573	0.589	0.715	0.610	0.622

Model	Zero-Shot Classes				All
Model	Lamp	Bed	Table	Sofa	All
Major.	0.333	0.333	0.333	0.333	0.333
*SG	0.501	0.564	0.637	0.536	0.560
PoE	0.422	0.466	0.587	0.483	0.490
NMN	0.462	0.492	0.572	0.532	0.515
MAC	0.533	0.531	0.632	0.551	0.567
NES
w/VGG16	0.544	0.578	0.693	0.588	0.601
w/Res101	0.573	0.589	0.715	0.610	0.622