imgutils.validate.dbrating

Overview:

A model for rating anime images into 4 classes (general, sensitive, questionable and explicit), based on danbooru rating system.

The following are sample images for testing.

The following are sample images for testing. (WARNING: NSFW!!!)

This is an overall benchmark of all the rating validation models:

../../_images/dbrating_benchmark.plot.py.svg

The models are hosted on huggingface - deepghs/anime_dbrating.

Note

This model is based on danbooru rating system, trained with 1.2 million images. If you need 3-level rating prediction, use imgutils.validate.rating.anime_rating().

Note

Please note that the classification of general, sensitive, questionable and explicit types does not have clear boundaries, making it challenging to clean the training data. As a result, there is no strict ground truth for the rating classification problem. The judgment functionality provided by the current module is intended as a quick and rough estimation.

If you require an accurate filtering or judgment function specifically for R-18 images, it is recommended to consider using object detection-based methods, such as using imgutils.detect.censor.detect_censors() to detect sensitive regions as the basis for judgment.

anime_dbrating_score

Overview:: Predict the rating of the given image, return the score with as a dict object.

Parameters:

image – Image to rating.
model_name – Model to use. Default is mobilenetv3_large_100_v0_ls0.2. All available models are listed on the benchmark plot above. If you need better accuracy, just set this to caformer_s36_v0_ls0.2.

Returns:

A dict with ratings and scores.

Examples::

>>> from imgutils.validate import anime_dbrating
>>>
>>> anime_dbrating('general/1.jpg')
('general', 0.7508869767189026)
>>> anime_dbrating('general/2.jpg')
('general', 0.7034655809402466)
>>> anime_dbrating('general/3.jpg')
('general', 0.728887677192688)
>>> anime_dbrating('general/4.jpg')
('general', 0.7404400110244751)
>>> anime_dbrating('sensitive/5.jpg')
('sensitive', 0.7446154952049255)
>>> anime_dbrating('sensitive/6.jpg')
('sensitive', 0.7514738440513611)
>>> anime_dbrating('sensitive/7.jpg')
('sensitive', 0.768704354763031)
>>> anime_dbrating('sensitive/8.jpg')
('sensitive', 0.8219676613807678)
>>> anime_dbrating('questionable/9.jpg')
('questionable', 0.7267540693283081)
>>> anime_dbrating('questionable/10.jpg')
('questionable', 0.7645740509033203)
>>> anime_dbrating('questionable/11.jpg')
('questionable', 0.7216582894325256)
>>> anime_dbrating('questionable/12.jpg')
('questionable', 0.7615436315536499)
>>> anime_dbrating('explicit/13.jpg')
('explicit', 0.815083920955658)
>>> anime_dbrating('explicit/14.jpg')
('explicit', 0.8321858644485474)
>>> anime_dbrating('explicit/15.jpg')
('explicit', 0.8204999566078186)
>>> anime_dbrating('explicit/16.jpg')
('explicit', 0.820833146572113)

anime_dbrating

Overview:: Predict the rating of the given image, return the class and its score.

Parameters:

image – Image to rating.
model_name – Model to use. Default is mobilenetv3_large_100_v0_ls0.2. All available models are listed on the benchmark plot above. If you need better accuracy, just set this to caformer_s36_v0_ls0.2.

Returns:

A tuple contains the rating and its score.

Examples::

>>> from imgutils.validate import anime_dbrating_score
>>>
>>> os.chdir('docs/source/api_doc/validate/dbrating')
>>>
>>> anime_dbrating_score('general/1.jpg')
{'general': 0.7508870363235474, 'sensitive': 0.11212056130170822, 'questionable': 0.06781744956970215, 'explicit': 0.06917501986026764}
>>> anime_dbrating_score('general/2.jpg')
{'general': 0.7034654021263123, 'sensitive': 0.15903906524181366, 'questionable': 0.06688199192285538, 'explicit': 0.07061357796192169}
>>> anime_dbrating_score('general/3.jpg')
{'general': 0.7288877964019775, 'sensitive': 0.1476859599351883, 'questionable': 0.060362350195646286, 'explicit': 0.06306383013725281}
>>> anime_dbrating_score('general/4.jpg')
{'general': 0.7404399514198303, 'sensitive': 0.10337048768997192, 'questionable': 0.08087948709726334, 'explicit': 0.07530999928712845}
>>> anime_dbrating_score('sensitive/5.jpg')
{'general': 0.055992450565099716, 'sensitive': 0.7446154356002808, 'questionable': 0.13191790878772736, 'explicit': 0.06747424602508545}
>>> anime_dbrating_score('sensitive/6.jpg')
{'general': 0.06458679586648941, 'sensitive': 0.7514738440513611, 'questionable': 0.10566363483667374, 'explicit': 0.07827574014663696}
>>> anime_dbrating_score('sensitive/7.jpg')
{'general': 0.07079866528511047, 'sensitive': 0.7687042951583862, 'questionable': 0.09974884241819382, 'explicit': 0.06074819341301918}
>>> anime_dbrating_score('sensitive/8.jpg')
{'general': 0.050435908138751984, 'sensitive': 0.8219675421714783, 'questionable': 0.0593985915184021, 'explicit': 0.06819795072078705}
>>> anime_dbrating_score('questionable/9.jpg')
{'general': 0.06569571048021317, 'sensitive': 0.1177448257803917, 'questionable': 0.726753830909729, 'explicit': 0.08980562537908554}
>>> anime_dbrating_score('questionable/10.jpg')
{'general': 0.06481882929801941, 'sensitive': 0.06922297924757004, 'questionable': 0.7645740509033203, 'explicit': 0.10138414055109024}
>>> anime_dbrating_score('questionable/11.jpg')
{'general': 0.06351721286773682, 'sensitive': 0.07683827728033066, 'questionable': 0.7216582894325256, 'explicit': 0.13798624277114868}
>>> anime_dbrating_score('questionable/12.jpg')
{'general': 0.05942752957344055, 'sensitive': 0.10584963858127594, 'questionable': 0.7615437507629395, 'explicit': 0.07317910343408585}
>>> anime_dbrating_score('explicit/13.jpg')
{'general': 0.060196295380592346, 'sensitive': 0.06751583516597748, 'questionable': 0.0572039857506752, 'explicit': 0.815083920955658}
>>> anime_dbrating_score('explicit/14.jpg')
{'general': 0.05398125201463699, 'sensitive': 0.06124086305499077, 'questionable': 0.0525919646024704, 'explicit': 0.8321859240531921}
>>> anime_dbrating_score('explicit/15.jpg')
{'general': 0.05922013148665428, 'sensitive': 0.06274889409542084, 'questionable': 0.057530902326107025, 'explicit': 0.8205001354217529}
>>> anime_dbrating_score('explicit/16.jpg')
{'general': 0.05683052912354469, 'sensitive': 0.06635929644107819, 'questionable': 0.05597696080803871, 'explicit': 0.8208332657814026}