Source code for catalyst.data.cv.mixins.blur

from typing import List
import random

import numpy as np

import albumentations as albu


class BlurMixin:
    """Calculates blur factor for augmented image."""

    def __init__(
        self,
        input_key: str = "image",
        output_key: str = "blur_factor",
        blur_min: int = 3,
        blur_max: int = 9,
        blur: List[str] = None,
    ):
        """
        Args:
            input_key: input key to use from annotation dict
            output_key: output key to use to store the result
        """
        self.input_key = input_key
        self.output_key = output_key

        self.blur_min = blur_min
        self.blur_max = blur_max
        blur = blur or ["Blur"]
        self.blur = [albu.__dict__[x]() for x in blur]
        self.num_blur = len(self.blur)
        self.num_blur_classes = blur_max - blur_min + 1 + 1
        self.blur_probability = (
            self.num_blur_classes - 1
        ) / self.num_blur_classes

    def __call__(self, dictionary):
        """@TODO: Docs. Contribution is welcome."""
        image = dictionary[self.input_key]
        blur_factor = 0

        if random.random() < self.blur_probability:
            blur_fn = np.random.choice(self.blur)
            blur_factor = int(
                np.random.randint(self.blur_min, self.blur_max)
                - self.blur_min
                + 1
            )
            image = blur_fn.apply(image=image, ksize=blur_factor)

        dictionary[self.input_key] = image
        dictionary[self.output_key] = blur_factor

        return dictionary


__all__ = ["BlurMixin"]