i_naturalist2021

• Description:

The iNaturalist dataset 2021 contains a total of 10,000 species. The full training dataset contains nearly 2.7M images. To make the dataset more accessible we have also created a "mini" training dataset with 50 examples per species for a total of 500K images. The full training train split overlaps with the mini split. The val set contains for each species 10 validation images (100K in total). There are a total of 500,000 test images in the public_test split (without ground-truth labels).

• Homepage: https://github.com/visipedia/inat_comp/tree/master/2021

• Source code: tfds.image_classification.i_naturalist2021.INaturalist2021

• Versions:

  • 1.0.0: Initial release.
  • 2.0.0: Update: Class indices follow the order in the JSON train file.
  • 2.0.1 (default): Update: Include the example id as provided in the JSON files.

• Download size: 316.54 GiB

• Dataset size: 318.45 GiB

• Auto-cached (documentation): No

• Splits:

Split	Examples
'mini'	500,000
'test'	500,000
'train'	2,686,843
'val'	100,000

• Feature structure:

FeaturesDict({
    'file_id': Text(shape=(), dtype=string),
    'id': Scalar(shape=(), dtype=int64),
    'image': Image(shape=(None, None, 3), dtype=uint8),
    'label': ClassLabel(shape=(), dtype=int64, num_classes=10000),
    'supercategory': ClassLabel(shape=(), dtype=int64, num_classes=11),
})

• Feature documentation:

Feature	Class	Shape	Dtype	Description
	FeaturesDict
file_id	Text		string
id	Scalar		int64
image	Image	(None, None, 3)	uint8
label	ClassLabel		int64
supercategory	ClassLabel		int64

• Supervised keys (See as_supervised doc): ('image', 'label')

• Figure (tfds.show_examples):

• Examples (tfds.as_dataframe):

• Citation:

\
@misc{inaturalist21,
    Howpublished = {~\url{https://github.com/visipedia/inat_comp/tree/master/2021} },
    Title = { {iNaturalist} 2021 competition dataset.},
    Year = {2021},
    key = { {iNaturalist} 2021 competition dataset},
    }