Skip to main content
Version: Next

ml.images.io

module ml.images.io

The io module provides standard helper functions to load images from disk

Global Variables#

  • IMREAD_COLOR

function load_image_from_disk#

load_image_from_disk(    path: str,    image_size=None,    convert_to_grey: bool = False,    keep_3d_shape=False) โ†’ <built-in function array>

Loads an image from file, applying preformatting

Args:

  • `path` (str): The filename of the image to load
  • `image_size` (tuple): The image size can be passed as tuple (W, H) or as int (W=H)
  • `convert_to_grey` (bool): This would reduce the size (and shape) of the image in making it a greyscale
  • `keep_3d_shape` (bool): Only used when convert_to_grey is true. Will keep the images in shape (H,W,1) in that case

Returns:

  • `np.array`: A numpy array that represents the image

function load_images#

load_images(    path: str,    image_size=None,    max_images: int = -1,    valid_extensions: <built-in function array> = ['.jpg', '.jpeg', '.gif', '.png'],    convert_to_grey: bool = False,    keep_3d_shape=False) โ†’ <built-in function array>

Loads the images from a specific folder

Args:

  • `path` (str): The path or folder name to load images from. This can be a relative or fully qualified path
  • `image_size` (tuple): The image size can be passed as tuple (W, H) or as int (W=H)
  • `max_images` (int): The maximum amount of images to load from the folder. If 0 or smaller, all images will be returned
  • `valid_extensions` (np.array): The file extensions that should be filtered. Defaults to jpg, jpeg, gif and png
  • `convert_to_grey` (bool): This would reduce the size (and shape) of the image in making it a greyscale
  • `keep_3d_shape` (bool): Only used when convert_to_grey is true. Will keep the images in shape (H,W,1) in that case

Returns:

  • `np.array`: A numpy array that contains all selected images represented as np.array

function load_image_from_url#

load_image_from_url(    image_url: str,    http_headers: dict = None,    image_size=None,    convert_to_grey: bool = False,    keep_3d_shape=False,    cache_location: str = None,    file_name: str = None,    force_download: bool = False) โ†’ <built-in function array>

Loads an image from a given url, applying preformatting and supporting file caching

Args:

  • `image_url` (str): The url to download the image.
  • `http_headers` (dict): The http headers to pass with the request as a dictionary
  • `image_size` (tuple): The image size can be passed as tuple (W, H) or as int (W=H)
  • `convert_to_grey` (bool): This would reduce the size (and shape) of the image in making it a greyscale
  • `keep_3d_shape` (bool): Only used when convert_to_grey is true. Will keep the images in shape (H,W,1) in that case
  • `cache_location` (str): When provided, the image will be cached to this folder location on disk
  • `file_name` (str): The file name of the image to be cached
  • `force_download` (bool): When true, the image will always be redownloaded and not retrieved from cache

Returns:

  • `np.array`: A numpy array that represents the image

function load_images_from_dataframe#

load_images_from_dataframe(    df: DataFrame,    image_column_name: str,    target_column_name: str,    image_size=None,    max_images: int = -1,    target_as_image: bool = False,    convert_to_grey: bool = False,    keep_3d_shape=False) โ†’ (<built-in function array>, <built-in function array>)

Loads a set images from disk, based on the file name in a Data Frame. And returns a related array (target) that can contain values from another column, or also images from disk

Args:

  • `df` (pd.DataFrame): the DataFrame, containing the references to the image
  • `image_column_name` (str): The name of the column that contains the image reference
  • `target_column_name` (str): The name of the column that contains the related target data
  • `image_size` (tuple): The image size can be passed as tuple (W, H) or as int (W=H)
  • `max_images` (int): The maximum amount of images to load from the folder. If 0 or smaller, all images will be returned
  • `target_as_image` (bool): Defines if the target column contains file names that should be loaded as image. If not, the column data will be used in the target array
  • `convert_to_grey` (bool): This would reduce the size (and shape) of the image in making it a greyscale
  • `keep_3d_shape` (bool): Only used when convert_to_grey is true. Will keep the images in shape (H,W,1) in that case

Returns: a tuple with the following objects:

  • `np.array`: A numpy array that contains all selected images represented as np.array
  • `np.array`: A numpy array that represents all targets that were asked

This file was automatically generated via lazydocs.