This dataset provides a plattform to benchmark transfer-learning algorithms, in particular attribute base classification [1,2]. 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 [3,4], thereby providing 85 numeric attribute values for each class. Using the shared attributes, it is possible to transfer information between different classes.
Attribute based classification and the data is described in:
 C. H. Lampert, H. Nickisch, and S. Harmeling. "Learning To Detect Unseen Object Classes by Between-Class Attribute Transfer". In CVPR, 2009 (pdf)
 C. H. Lampert, H. Nickisch, and S. Harmeling. "Attribute-Based Classification for Zero-Shot Visual Object Categorization". IEEE T-PAMI, 2013 (pdf)
The class/attribute matrix was originally created by:
 D. N. Osherson, J. Stern, O. Wilkie, M. Stob, and E. E. Smith. "Default probability". Cognitive Science, 15(2), 1991.
 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.
Version 1.0, May 13th 2009
Because the dataset is rather large, the download is split into separate archives:
Addendum: new attributes.py script that work with recent versions of Shogun
Attribute based classification results (class-averaged multiclass accuracy):
- direct-attribute prediction (DAP): 40.5%  41.4% 
- indirect-attribute prediction (IAP): 27.8%  42.2% 
- Attribute/Hierchical Label Embedding (AHLE): 43.5%; 
- Archive with confusion matrices and precision-recall curves AwA-results.tar
Unpublished results (latest update: 21/07/2015):
|method||features/kernel||multiclass acc||mean class AUC||mean attributes AUC||comments|
|DAP||DECAF (d=4096)/linear||47.7%||83.1 ± 8.3||0.73 ± 0.14||default split|
|DAP||VGG19 (d=40960)/linear||57.5%||86.3 ± 9.5||0.78 ± 0.14||default split|
more results to follow...