metadata

task_categories:
  - image-classification

AutoTrain Dataset for project: vit-skin-derna

Dataset Description

This dataset has been automatically processed by AutoTrain for project vit-skin-derna.

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": "<32x32 RGB PIL image>",
    "target": 4
  },
  {
    "image": "<32x32 RGB PIL image>",
    "target": 8
  }
]

Dataset Fields

The dataset has the following fields (also called "features"):

{
  "image": "Image(decode=True, id=None)",
  "target": "ClassLabel(names=['airplane', 'automobile', 'bird', 'cat', 'deer', 'dog', 'frog', 'horse', 'ship', 'truck'], 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	40000
valid	10000