"""
Overview:
A model for screening monochrome images, with the definition of monochrome images referring to Danbooru.
The following are testing images. The top two rows are monochrome images, and the bottom two rows are color images.
Please note that **monochrome images are not only those with all pixels in grayscale**.
.. image:: monochrome.plot.py.svg
:align: center
This is an overall benchmark of all the monochrome validation models:
.. image:: monochrome_benchmark.plot.py.svg
:align: center
The models are hosted on `huggingface - deepghs/monochrome_detect <https://huggingface.co/deepghs/monochrome_detect>`_.
"""
from ..data import ImageTyping
from ..generic import classify_predict_score, classify_predict
__all__ = [
'get_monochrome_score',
'is_monochrome',
]
_DEFAULT_MODEL_NAME = 'mobilenetv3_large_100_dist_safe2'
_REPO_ID = 'deepghs/monochrome_detect'
[docs]def get_monochrome_score(image: ImageTyping, model_name: str = _DEFAULT_MODEL_NAME) -> float:
"""
Overview:
Get monochrome score of the given image.
:param image: Image to predict, can be a ``PIL.Image`` object or the path of the image file.
:param model_name: The model used for inference. The default value is ``mobilenetv3_dist``,
which offers high runtime performance. If you need better accuracy, just use ``caformer_s36``.
Examples::
>>> from imgutils.validate import get_monochrome_score
>>>
>>> get_monochrome_score('mono/1.jpg') # monochrome images
0.9614395499229431
>>> get_monochrome_score('mono/2.jpg')
0.9458909034729004
>>> get_monochrome_score('mono/3.jpg')
0.9559807777404785
>>> get_monochrome_score('mono/4.jpg')
0.9651952981948853
>>> get_monochrome_score('mono/5.jpg')
0.9379720687866211
>>> get_monochrome_score('mono/6.jpg')
0.8814834356307983
>>>
>>> get_monochrome_score('colored/7.jpg') # colored images
0.03941023349761963
>>> get_monochrome_score('colored/8.jpg')
0.07492382079362869
>>> get_monochrome_score('colored/9.jpg')
0.09546589106321335
>>> get_monochrome_score('colored/10.jpg')
0.016521310433745384
>>> get_monochrome_score('colored/11.jpg')
0.005693843588232994
>>> get_monochrome_score('colored/12.jpg')
0.0315730981528759
"""
return classify_predict_score(image, _REPO_ID, model_name)['monochrome']
[docs]def is_monochrome(image: ImageTyping, threshold: float = 0.5,
model_name: str = _DEFAULT_MODEL_NAME) -> bool:
"""
Overview:
Predict if the image is monochrome.
:param image: Image to predict, can be a ``PIL.Image`` object or the path of the image file.
:param threshold: Threshold value during prediction. If the score is higher than the threshold,
the image will be classified as monochrome.
:param model_name: The model used for inference. The default value is ``mobilenetv3_dist``,
which offers high runtime performance. If you need better accuracy, just use ``caformer_s36``.
Examples:
>>> import os
>>> from imgutils.validate import is_monochrome
>>>
>>> is_monochrome('mono/1.jpg') # monochrome images
True
>>> is_monochrome('mono/2.jpg')
True
>>> is_monochrome('mono/3.jpg')
True
>>> is_monochrome('mono/4.jpg')
True
>>> is_monochrome('mono/5.jpg')
True
>>> is_monochrome('mono/6.jpg')
True
>>> is_monochrome('colored/7.jpg') # colored images
False
>>> is_monochrome('colored/8.jpg')
False
>>> is_monochrome('colored/9.jpg')
False
>>> is_monochrome('colored/10.jpg')
False
>>> is_monochrome('colored/11.jpg')
False
>>> is_monochrome('colored/12.jpg')
False
"""
type_, _ = classify_predict(image, _REPO_ID, model_name)
return type_ == 'monochrome'