Utils¶
A set of utility functions.
All of the functions can be imported from kiui
directly.
Examples¶
import kiui
x = kiui.read_image('image.png')
kiui.lo(x)
API¶
- kiui.utils.lo(*xs, verbose=0)[source]¶
inspect array like objects and report statistics.
- Parameters:
xs (Any) – array like objects to inspect.
verbose (int, optional) – level of verbosity, set to 1 to report mean and std, 2 to print the content. Defaults to 0.
- kiui.utils.seed_everything(seed=42, verbose=False, strict=False)[source]¶
auto set seed for random, numpy and torch.
- Parameters:
seed (int, optional) – random seed. Defaults to 42.
verbose (bool, optional) – whether to report each seed setting. Defaults to False.
strict (bool, optional) – whether to use strict deterministic mode for better torch reproduction. Defaults to False.
- kiui.utils.read_json(path)[source]¶
load a json file.
- Parameters:
path (str) – path to json file.
- Returns:
json content.
- Return type:
dict
- kiui.utils.write_json(path, x)[source]¶
write a json file.
- Parameters:
path (str) – path to write json file.
x (dict) – dict to write.
- kiui.utils.read_pickle(path)[source]¶
read a pickle file.
- Parameters:
path (str) – path to pickle file.
- Returns:
pickle content.
- Return type:
Any
- kiui.utils.write_pickle(path, x)[source]¶
write a pickle file.
- Parameters:
path (str) – path to write pickle file.
x (Any) – content to write.
- kiui.utils.read_image(path: str, mode: Literal[‘float’, ‘uint8’, ‘pil’, ‘torch’, ‘tensor’] = 'float', order: Literal[‘RGB’, ‘RGBA’, ‘BGR’, ‘BGRA’] = 'RGB')[source]¶
read an image file into various formats and color mode.
- Parameters:
path (str) – path to the image file.
mode (Literal[“float”, “uint8”, “pil”, “torch”, “tensor”], optional) – returned image format. Defaults to “float”. float: float32 numpy array, range [0, 1]; uint8: uint8 numpy array, range [0, 255]; pil: PIL image; torch/tensor: float32 torch tensor, range [0, 1];
order (Literal[“RGB”, “RGBA”, “BGR”, “BGRA”], optional) – channel order. Defaults to “RGB”.
Note
By default this function will convert RGBA image to white-background RGB image. Use
order="RGBA"
to keep the alpha channel.- Returns:
the image array.
- Return type:
Union[np.ndarray, PIL.Image, torch.Tensor]
- kiui.utils.write_image(path: str, img: Tensor | ndarray | Image, order: Literal[‘RGB’, ‘BGR’] = 'RGB')[source]¶
write an image to various formats.
- Parameters:
path (str) – path to write the image file.
img (Union[torch.Tensor, np.ndarray, PIL.Image.Image]) – image to write.
order (str, optional) – channel order. Defaults to “RGB”.
- kiui.utils.load_file_from_url(url, model_dir=None, progress=True, file_name=None)[source]¶
Load file form http url, will download models if necessary.
- Parameters:
url (str) – URL to be downloaded.
model_dir (str) – The path to save the downloaded model. Should be a full path. If None, use pytorch hub_dir. Default: None.
progress (bool) – Whether to show the download progress. Default: True.
file_name (str) – The downloaded file name. If None, use the file name in the url. Default: None.
- Returns:
The path to the downloaded file.
- Return type:
str
- kiui.utils.is_format(f: str, format: Sequence[str])[source]¶
if a file’s extension is in a set of format
- Parameters:
f (str) – file name.
format (Sequence[str]) – set of extensions (both ‘.jpg’ or ‘jpg’ is ok).
- Returns:
if the file’s extension is in the set.
- Return type:
bool
- kiui.utils.batch_process_files(process_fn, path, out_path, overwrite=False, in_format=['.jpg', '.jpeg', '.png'], out_format=None, image_mode='uint8', image_color_order='RGB', **kwargs)[source]¶
simple function wrapper to batch processing files.
- Parameters:
process_fn (Callable) – process function.
path (str) – path to a file or a directory containing the files to process.
out_path (str) – output path of a file or a directory.
overwrite (bool, optional) – whether to overwrite existing results. Defaults to False.
in_format (list, optional) – input file formats. Defaults to [“.jpg”, “.jpeg”, “.png”].
out_format (str, optional) – output file format. Defaults to None.
image_mode (str, optional) – for images, the mode to read. Defaults to ‘uint8’.
image_color_order (str, optional) – for images, the color order. Defaults to “RGB”.