Label-Embedding for Image Classification

Attributes act as intermediate representations that enable parameter sharing between classes, a must when training data is scarce. We propose to view attribute-based image classification as a label-embedding problem: each class is embedded in the space of attribute vectors. We introduce a function t...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:IEEE transactions on pattern analysis and machine intelligence Jg. 38; H. 7; S. 1425 - 1438
Hauptverfasser: Akata, Zeynep, Perronnin, Florent, Harchaoui, Zaid, Schmid, Cordelia
Format: Journal Article
Sprache:Englisch
Veröffentlicht: United States IEEE 01.07.2016
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Institute of Electrical and Electronics Engineers
Schlagworte:
Animals
Attributes
Computer Science
Computer vision
Computer Vision and Pattern Recognition
Encoding
Fine Grained Image Classification
Image classification
Label Embedding
Linear programming
Loss measurement
Subspace Learning
Taxonomy
Training
attributes
label embedding
zero-shot learning
Image classification
Image Classification
Attributes
Label Embedding
Zero-Shot Learning
ISSN:0162-8828, 2160-9292, 1939-3539
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Attributes act as intermediate representations that enable parameter sharing between classes, a must when training data is scarce. We propose to view attribute-based image classification as a label-embedding problem: each class is embedded in the space of attribute vectors. We introduce a function that measures the compatibility between an image and a label embedding. The parameters of this function are learned on a training set of labeled samples to ensure that, given an image, the correct classes rank higher than the incorrect ones. Results on the Animals With Attributes and Caltech-UCSD-Birds datasets show that the proposed framework outperforms the standard Direct Attribute Prediction baseline in a zero-shot learning scenario. Label embedding enjoys a built-in ability to leverage alternative sources of information instead of or in addition to attributes, such as, e.g., class hierarchies or textual descriptions. Moreover, label embedding encompasses the whole range of learning settings from zero-shot learning to regular learning with a large number of labeled examples.
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
content type line 23
ISSN:0162-8828
2160-9292
1939-3539
DOI:10.1109/TPAMI.2015.2487986