Large Scale Metric Learning for Distance-Based Image Classification - Archive ouverte HAL Access content directly
Reports (Research Report) Year : 2012

Large Scale Metric Learning for Distance-Based Image Classification

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

Abstract

This paper studies large-scale image classification, in a setting where new classes and training images could continuously be added at (near) zero cost. We cast this problem into one of learning a low-rank metric, which is shared across all classes and explore k-nearest neighbor (k-NN) and nearest class mean (NCM) classifiers. We also introduce an extension of the NCM classifier to allow for richer image representations. Experiments on the ImageNet 2010 challenge dataset ---which contains more than 1M training images of 1K classes--- shows, surprisingly, that the NCM classifier compares favorably to the more flexible k-NN classifier. Moreover, the NCM performance suggests that 256 dimensional features is comparable to that of linear SVMs, which were used to obtain the current state-of-the-art performance. Experimentally we study the generalization performance to classes that were not used to learn the metrics, and show how a zero-shot model based on the ImageNet hierarchy can be combined effectively with small training datasets. Using a metric learned on 1K classes, we show results for the ImageNet-10K dataset, and obtain performance that is competitive with the current state-of-the-art, while requiring significant less training time.
Fichier principal
Vignette du fichier
RR-8077.pdf (2.79 Mo) Télécharger le fichier
Vignette du fichier
RR-8077.web.png (18.94 Ko) Télécharger le fichier
Origin : Explicit agreement for this submission
Format : Figure, Image
Loading...

Dates and versions

hal-00735908 , version 1 (27-09-2012)

Identifiers

  • HAL Id : hal-00735908 , version 1

Cite

Thomas Mensink, Jakob Verbeek, Florent Perronnin, Gabriela Csurka. Large Scale Metric Learning for Distance-Based Image Classification. [Research Report] RR-8077, INRIA. 2012, pp.30. ⟨hal-00735908⟩
995 View
600 Download

Share

Gmail Facebook Twitter LinkedIn More