Documentation for image_transforms

Functionality

The image_transforms function applies a transformation to images stored in a dataset. It calls the provided transform function, which defaults to center_crop_transform, to generate square images of specified size n_pixels. If the parameter del_images is set to True, the original images will be removed from the dataset.

Parameters

• examples: A dataset containing the images.
• transform: A callable transformation function. Default is center_crop_transform.
• n_pixels: Size (in pixels) of the square image to produce.
• image_field_name: The field name containing the original images.
• pixel_values_name: The field to store the transformed images.
• del_images: A Boolean flag to indicate whether to delete the original images.

Usage

The purpose of the image_transforms function is to process image data in a dataset by applying a customizable transformation.

Example

result_dataset = image_transforms(
    examples=dataset,
    transform=center_crop_transform,
    n_pixels=224,
    image_field_name="item",
    pixel_values_name="pixel_values",
    del_images=False
)