imgutils.validate.safe

Overview:

Check if the images are polluted or safe.

This is an overall benchmark of all the safe check models:

../../_images/safe_benchmark.plot.py.svg

Inspired from mf666/shit-checker.

safe_check_score

imgutils.validate.safe.safe_check_score(image: str | PathLike | bytes | bytearray | BinaryIO | Image, model_name: str = 'mobilenet.xs.v2', max_batch_size: int = 8) → Mapping[str, float][source]

Check the safety score of an image.

Parameters:

  • image (ImageTyping) – The image to check.

  • model_name (str) – The name of the safety model.

  • max_batch_size (int) – The maximum batch size for prediction.

Returns:

A mapping of safety labels and their corresponding scores.

Return type:

Mapping[str, float]

safe_check

imgutils.validate.safe.safe_check(image: str | PathLike | bytes | bytearray | BinaryIO | Image, model_name: str = 'mobilenet.xs.v2', max_batch_size: int = 8) → Tuple[str, float][source]

Check the safety label and score of an image.

Parameters:

  • image (ImageTyping) – The image to check.

  • model_name (str) – The name of the safety model.

  • max_batch_size (int) – The maximum batch size for prediction.

Returns:

A tuple containing the safety label and score.

Return type:

Tuple[str, float]