dev(narugo): add new benchmark

import random

from benchmark import BaseBenchmark, create_plot_cli

from imgutils.restore.scunet import SCUNetModelTyping, restore_with_scunet

class SCUNetBenchmark(BaseBenchmark):

    def __init__(self, model: SCUNetModelTyping):

        BaseBenchmark.__init__(self)

        self.model = model

    def load(self):

        from imgutils.restore.scunet import _open_scunet_model

        _open_scunet_model(self.model)

    def unload(self):

        from imgutils.restore.scunet import _open_scunet_model

        _open_scunet_model.cache_clear()

    def run(self):

        image_file = random.choice(self.all_images)

        _ = restore_with_scunet(image_file, model=self.model)

if __name__ == '__main__':

    create_plot_cli(

        [

            ('SCUNet-GAN', SCUNetBenchmark('GAN')),

            ('SCUNet-PSNR', SCUNetBenchmark('PSNR')),

        ],

        title='Benchmark for SCUNet Models',

        run_times=5,

        try_times=10,

    )()

"""

Overview:

    Restore the images using `NafNet <https://github.com/megvii-research/NAFNet>`_.

    .. image:: nafnet_demo.plot.py.svg

        :align: center

    This is an overall benchmark of all the NafNet models:

    .. image:: nafnet_benchmark.plot.py.svg

        :align: center

    .. warning::

        Currently, we've identified a significant issue with NafNet when images contain gaussian noise.

        To ensure your code functions correctly, please ensure the credibility of

        your image source or preprocess them using SCUNet.

"""

from functools import lru_cache

from typing import Literal

"""

Overview:

    Restore the images using `SCUNet <https://github.com/cszn/SCUNet>`_.

    .. image:: scunet_demo.plot.py.svg

        :align: center

    This is an overall benchmark of all the SCUNet models:

    .. image:: scunet_benchmark.plot.py.svg

        :align: center

"""

from functools import lru_cache

from typing import Literal