Large-Scale Zero-Shot Image Classification from Rich and Diverse Textual Descriptions

Sebastian Bujwid, Josephine Sullivan


Abstract
We study the impact of using rich and diverse textual descriptions of classes for zero-shot learning (ZSL) on ImageNet. We create a new dataset ImageNet-Wiki that matches each ImageNet class to its corresponding Wikipedia article. We show that merely employing these Wikipedia articles as class descriptions yields much higher ZSL performance than prior works. Even a simple model using this type of auxiliary data outperforms state-of-the-art models that rely on standard features of word embedding encodings of class names. These results highlight the usefulness and importance of textual descriptions for ZSL, as well as the relative importance of auxiliary data type compared to the algorithmic progress. Our experimental results also show that standard zero-shot learning approaches generalize poorly across categories of classes.
Anthology ID:
2021.lantern-1.4
Volume:
Proceedings of the Third Workshop on Beyond Vision and LANguage: inTEgrating Real-world kNowledge (LANTERN)
Month:
April
Year:
2021
Address:
Kyiv, Ukraine
Editors:
Marius Mosbach, Michael A. Hedderich, Sandro Pezzelle, Aditya Mogadala, Dietrich Klakow, Marie-Francine Moens, Zeynep Akata
Venue:
LANTERN
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
38–52
Language:
URL:
https://aclanthology.org/2021.lantern-1.4
DOI:
Bibkey:
Cite (ACL):
Sebastian Bujwid and Josephine Sullivan. 2021. Large-Scale Zero-Shot Image Classification from Rich and Diverse Textual Descriptions. In Proceedings of the Third Workshop on Beyond Vision and LANguage: inTEgrating Real-world kNowledge (LANTERN), pages 38–52, Kyiv, Ukraine. Association for Computational Linguistics.
Cite (Informal):
Large-Scale Zero-Shot Image Classification from Rich and Diverse Textual Descriptions (Bujwid & Sullivan, LANTERN 2021)
Copy Citation:
PDF:
https://aclanthology.org/2021.lantern-1.4.pdf
Supplementary material:
 2021.lantern-1.4.Supplementary_material.zip
Data
AwA2