Metric Learning for Large Scale Image Classification: Generalizing to New Classes at Near-Zero Cost - Archive ouverte HAL Access content directly
Conference Papers Year : 2012

Metric Learning for Large Scale Image Classification: Generalizing to New Classes at Near-Zero Cost

(1, 2) , (1) , (2) , (2)
1
2

Abstract

We are interested in large-scale image classification and especially in the setting where images corresponding to new or existing classes are continuously added to the training set. Our goal is to devise classifiers which can incorporate such images and classes on-the-fly at (near) zero cost. We cast this problem into one of learning a metric which is shared across all classes and explore k-nearest neighbor (k-NN) and nearest class mean (NCM) classifiers. We learn metrics on the ImageNet 2010 challenge data set, which contains more than 1.2M training images of 1K classes. Surprisingly, the NCM classifier compares favorably to the more flexible k-NN classifier, and has comparable performance to linear SVMs. We also study the generalization performance, among others by using the learned metric on the ImageNet-10K dataset, and we obtain competitive performance. Finally, we explore zero-shot classification, and show how the zero-shot model can be combined very effectively with small training datasets.
Vignette du fichier
classmeansfig.web.png (537.74 Ko) Télécharger le fichier Fichier principal
Vignette du fichier
mensink12eccv.final.pdf (879.53 Ko) Télécharger le fichier
Format : Figure, Image
Origin : Files produced by the author(s)
Loading...

Dates and versions

hal-00722313 , version 1 (01-08-2012)

Identifiers

Cite

Thomas Mensink, Jakob Verbeek, Florent Perronnin, Gabriela Csurka. Metric Learning for Large Scale Image Classification: Generalizing to New Classes at Near-Zero Cost. ECCV 2012 - 12th European Conference on Computer Vision, Oct 2012, Florence, Italy. pp.488-501, ⟨10.1007/978-3-642-33709-3_35⟩. ⟨hal-00722313⟩
1086 View
9486 Download

Altmetric

Share

Gmail Facebook Twitter LinkedIn More