Animals on the Web

Our dataset, Animals on the Web, consists of images from 10 animal categories collected from the web using Google Text Search on the category name. In addition one category, monkey, has been extended to a much larger set of images using monkey plus related monkey terms. These images were collected for the system described in the paper: "Animals on the Web", CVPR '06.

We have provided both the positive and negative images from each search along with ground truth labels. These labels are somewhat arbitrary as deciding what belongs to a category is subjective. The labels may also contain some errors.

Animals on the Web

To unpack the files into your current directory under unix use the command: tar zxvf category.tar.gz

The download directory contains one tar.gz file per category. "othermonkeys" indicates the large extended monkey class

Each file will unzip into a directory containing:

  • the images from that category,
  • a matfile called animaldata_category.mat (where category = {alligator,ant,bear,beaver,dolphin,frog,giraffe,leopard,monkey, penguin}),
  • two html files showing the positive samples (possamples.html) and negative samples (negsamples.html) for that category.
  • The matfiles animaldata_category.mat contain two variables, imgnames which gives the picture filenames and gt where gt(i) gives the ground truth label for image imgnames{i} (1 indicates positive, 0 negative).


    This dataset is for academic research purposes only. If you use our dataset please reference:

  • Animals on the Web [pdf] [ps]
    Tamara L. Berg, David A. Forsyth
    Computer Vision and Pattern Recognition (CVPR), 2006

  • Just for fun -- Mean images for each category

    Alligator Ant Bear Beaver Dolphin
    Frog Giraffe Leopard Monkey Penguin
    Large Monkey