Disclaimer

The Animals with Attributes dataset is currently suspended. Its images are not available anymore because of copyright restrictions. Please use the drop-in replacement, Animals with Attributes 2.


Overview

This dataset provides a plattform to benchmark transfer-learning algorithms, in particular attribute base classification [1]. It consists of 30475 images of 50 animals classes with six pre-extracted feature representations for each image. The animals classes are aligned with Osherson's classical class/attribute matrix [2,3], thereby providing 85 numeric attribute values for each class. Using the shared attributes, it is possible to transfer information between different classes.

Publication

Attribute based classification and the data is described in:

[1] C. H. Lampert, H. Nickisch, and S. Harmeling. "Learning To Detect Unseen Object Classes by Between-Class Attribute Transfer". In CVPR, 2009 (pdf)

The class/attribute matrix was originally created by:

[2] D. N. Osherson, J. Stern, O. Wilkie, M. Stob, and E. E. Smith. "Default probability". Cognitive Science, 15(2), 1991.

[3] C. Kemp, J. B. Tenenbaum, T. L. Griffiths, T. Yamada, and N. Ueda. "Learning systems of concepts with an infinite relational model". In AAAI, 2006.

Downloads

Version 1.0, May 13th 2009

Because the dataset is rather large, the download is split into separate archives:

  • Base package (1M) including the class/attribute table:AwA-base.tar.bz2 (everybody needs this)
  • Color Histogram features (124M): AwA-features-cq.tar.bz2
  • Local Self-Similarity features (30M): AwA-features-lss.tar.bz2
  • PyramidHOG (PHOG) features (28M): AwA-features-phog.tar.bz2
  • SIFT features (44M): AwA-features-sift.tar.bz2
  • colorSIFT features (44M): AwA-features-rgsift.tar.bz2
  • SURF features (49M): AwA-features-surf.tar.bz2
  • Source code (30K) illustrating DAP and IAP methods: AwA-code.tar.bz2
    Addendum: new attributes.py script that work with recent versions of Shogun
  • Full Image Set in JPEG format: not available for copyright reasons
    please ask at <chl(at)ist.ac.at>.
  • Results

    Attribute based classification results (class-averaged multiclass accuracy):

  • direct-attribute prediction (DAP) 40.5%;       [1]
  • indirect-attribute prediction (IAP) 27.8%;       [1]
  • Archive with confusion matrices and precision-recall curves AwA-results.tar
  • more results to follow...