metadata
task_categories:
- image-classification
AutoTrain Dataset for project: ethnicity-test_v003
Dataset Description
This dataset has been automatically processed by AutoTrain for project ethnicity-test_v003.
Languages
The BCP-47 code for the dataset's language is unk.
Dataset Structure
Data Instances
A sample from this dataset looks as follows:
[
{
"image": "<512x512 RGB PIL image>",
"target": 1
},
{
"image": "<512x512 RGB PIL image>",
"target": 3
}
]
Dataset Fields
The dataset has the following fields (also called "features"):
{
"image": "Image(decode=True, id=None)",
"target": "ClassLabel(names=['african', 'asian', 'caucasian', 'hispanic', 'indian'], id=None)"
}
Dataset Splits
This dataset is split into a train and validation split. The split sizes are as follow:
| Split name | Num samples |
|---|---|
| train | 4531 |
| valid | 1135 |