sbu_captions

References:

Use the following command to load this dataset in TFDS:

ds = tfds.load('huggingface:sbu_captions')
  • Description:
The SBU Captioned Photo Dataset is a collection of over 1 million images with associated text descriptions extracted from Flicker.
  • License: unknown
  • Version: 0.0.0
  • Splits:
Split Examples
'train' 1000000
  • Features:
{
    "image_url": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "user_id": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    },
    "caption": {
        "dtype": "string",
        "id": null,
        "_type": "Value"
    }
}