sdeval.corrupt.aicorrupt

Overview:

AI image corrupt evaluation metrics.

get_ai_corrupted

sdeval.corrupt.aicorrupt.get_ai_corrupted(image: Union[str, os.PathLike, bytes, bytearray, BinaryIO, PIL.Image.Image], model_name: str = 'caformer_s36_v0_focal') → Mapping[str, float][source]

Get AI image corrupted detection scores for an image.

This function calculates AI image corrupted detection scores for a given image using the specified model.

Parameters:

  • image (ImageTyping) – The input image.

  • model_name (str) – The name of the AI image corrupted detection model. Default is ‘caformer_s36_v0_focal’.

Returns:

A dictionary containing the corrupted score.

Return type:

Mapping[str, float]

AICorruptMetrics

class sdeval.corrupt.aicorrupt.AICorruptMetrics(model_name: str = 'caformer_s36_v0_focal', silent: bool = False, tqdm_desc: str = None)[source]

Class for calculating an AI image corruptness score.

The AICorruptMetrics class allows you to calculate an AI image corruptness score using the AI image corrupted detection model.

Parameters:

  • model_name (str) – The name of the AI image corrupted detection model. Default is ‘caformer_s36_v0_focal’.

  • silent (bool) – If True, suppresses progress bars and additional output during calculation.

  • tqdm_desc (str) – Description for the tqdm progress bar during calculation.

__init__(model_name: str = 'caformer_s36_v0_focal', silent: bool = False, tqdm_desc: str = None)[source]

Initialize self. See help(type(self)) for accurate signature.

score(images: Union[PIL.Image.Image, str, List[Union[PIL.Image.Image, str]]], silent: bool = None, mode: Literal[mean, seq] = 'mean') → Union[float, numpy.ndarray][source]

Calculate the AI image corruptness score for a set of images.

This method calculates the AI image corruptness score for a set of input images using the AI image corrupted detection model.

Parameters:

  • images (ImagesTyping) – The set of input images for calculating the AI image corruptness score.

  • silent (bool) – If True, suppresses progress bars and additional output during calculation.

  • mode (Literal['mean', 'seq']) – Mode of the return value. Return a float value when mean is assigned, return a numpy array when seq is assigned. Default is mean.

Returns:

The AI image corruptness score.

Return type:

Union[float, np.ndarray]