maurice-fp/stanford-dogs

Dataset Card for Stanford Dogs

The Stanford Dogs dataset contains images of 120 breeds of dogs from around the world. This dataset has been built using images and annotation from ImageNet for the task of fine-grained image categorization. Contents of this dataset:

• Number of categories: 120

• Number of images: 20,580

• Annotations: Class labels, Bounding boxes (not imported to HF)

• Website: http://vision.stanford.edu/aditya86/ImageNetDogs/

• Paper: http://people.csail.mit.edu/khosla/papers/fgvc2011.pdf

Citation

@inproceedings{KhoslaYaoJayadevaprakashFeiFei_FGVC2011,
  author = "Aditya Khosla and Nityananda Jayadevaprakash and Bangpeng Yao and Li Fei-Fei",
  title = "Novel Dataset for Fine-Grained Image Categorization",
  booktitle = "First Workshop on Fine-Grained Visual Categorization, IEEE Conference on Computer Vision and Pattern Recognition",
  year = "2011",
  month = "June",
  address = "Colorado Springs, CO",
}