plugins/train/trainer/trainer_config.py - master - githubmirror/faceswap

plugins/train/trainer/trainer_config.py
175 строк6 KB
torzdf
Train: Migrate data loading to Torch (#1540)
16 апр 2026, 21:19
16 апр 2026, 21:1983fdd5f
#!/usr/bin/env python3
""" Default configurations for trainers """
from __future__ import annotations

import inspect
import gettext
import logging
import typing as T

from dataclasses import dataclass

from lib.config import ConfigItem, GlobalSection
from lib.utils import get_module_objects

logger = logging.getLogger(__name__)


# LOCALES
_LANG = gettext.translation("plugins.train.trainer.train_config",
                            localedir="locales", fallback=True)
_ = _LANG.gettext


@dataclass
class Loader(GlobalSection):
    """ trainer.loader section"""

    helptext = _(
        "Data Loader Options.\n"
        "Controls how training data is loaded from disk")

    num_processes = ConfigItem(
        datatype=int,
        default=4,
        group=_("data loading"),
        info=_("Number of processors to use for loading and processing data from disk. 0 to just "
               "use the Main process."),
        rounding=1,
        min_max=(0, 32))

    pre_fetch = ConfigItem(
        datatype=int,
        default=2,
        group=_("data loading"),
        info=_("The Number of items that each loader should pre-fetch and hold in RAM. Default is "
               "usually fine unless you have disk contention with variable read speeds."),
        rounding=1,
        min_max=(1, 10))


@dataclass
class Augmentation(GlobalSection):
    """ trainer.augmentation section"""

    helptext = _(
        "Data Augmentation Options.\n"
        "WARNING: The defaults for augmentation will be fine for 99.9% of use cases. "
        "Only change them if you absolutely know what you are doing!")

    preview_images = ConfigItem(
        datatype=int,
        default=14,
        group=_("evaluation"),
        info=_("Number of sample faces to display for each side in the preview when training."),
        rounding=2,
        min_max=(2, 16))

    mask_opacity = ConfigItem(
        datatype=int,
        default=30,
        group=_("evaluation"),
        info=_("The opacity of the mask overlay in the training preview. Lower values are more "
               "transparent."),
        rounding=2,
        min_max=(0, 100))

    mask_color = ConfigItem(
        datatype=str,
        default="#ff0000",
        choices="colorchooser",
        group=_("evaluation"),
        info=_("The RGB hex color to use for the mask overlay in the training preview."))

    zoom_amount = ConfigItem(
        datatype=int,
        default=5,
        group=_("image augmentation"),
        info=_("Percentage amount to randomly zoom each training image in and out."),
        rounding=1,
        min_max=(0, 25))

    rotation_range = ConfigItem(
        datatype=int,
        default=10,
        group=_("image augmentation"),
        info=_("Percentage amount to randomly rotate each training image."),
        rounding=1,
        min_max=(0, 25))

    shift_range = ConfigItem(
        datatype=int,
        default=5,
        group=_("image augmentation"),
        info=_("Percentage amount to randomly shift each training image horizontally and "
               "vertically."),
        rounding=1,
        min_max=(0, 25))

    flip_chance = ConfigItem(
        datatype=int,
        default=50,
        group=_("image augmentation"),
        info=_("Percentage chance to randomly flip each training image horizontally.\n"
               "NB: This is ignored if the 'no-flip' option is enabled"),
        rounding=1,
        min_max=(0, 75))

    color_lightness = ConfigItem(
        datatype=int,
        default=30,
        group=_("color augmentation"),
        info=_("Percentage amount to randomly alter the lightness of each training image.\n"
               "NB: This is ignored if the 'no-augment-color' option is enabled"),
        rounding=1,
        min_max=(0, 75))

    color_ab = ConfigItem(
        datatype=int,
        default=8,
        group=_("color augmentation"),
        info=_("Percentage amount to randomly alter the 'a' and 'b' colors of the L*a*b* color "
               "space of each training image.\nNB: This is ignored if the 'no-augment-color' "
               "option is enabled"),
        rounding=1,
        min_max=(0, 50))

    color_clahe_chance = ConfigItem(
        datatype=int,
        default=50,
        group=_("color augmentation"),
        info=_("Percentage chance to perform Contrast Limited Adaptive Histogram Equalization on "
               "each training image.\nNB: This is ignored if the 'no-augment-color' option is "
               "enabled"),
        rounding=1,
        min_max=(0, 75),
        fixed=False)

    color_clahe_max_size = ConfigItem(
        datatype=int,
        default=4,
        group=_("color augmentation"),
        info=_("The grid size dictates how much Contrast Limited Adaptive Histogram Equalization "
               "is performed on any training image selected for clahe. Contrast will be applied "
               "randomly with a grid-size of 0 up to the maximum. This value is a multiplier "
               "calculated from the training image size.\nNB: This is ignored if the "
               "'no-augment-color' option is enabled"),
        rounding=1,
        min_max=(1, 8))


def get_defaults() -> dict[str, GlobalSection]:
    """ Obtain the default values for adding to the config.ini file

    Returns
    -------
    defaults
        The option names and config items
    """
    defaults = {k: T.cast(GlobalSection, v) for k, v in globals().items()
                if inspect.isclass(v) and issubclass(v, GlobalSection) and v != GlobalSection}
    logger.debug("Training config. options: %s", defaults)
    return defaults


__all__ = get_module_objects(__name__)