mobilenetv3.py

import warnings
from functools import partial
from typing import Any, Callable, List, Optional, Sequence

import torch
from torch import nn, Tensor

from ..ops.misc import Conv2dNormActivation, SqueezeExcitation as SElayer
from ..transforms._presets import ImageClassification
from ..utils import _log_api_usage_once
from ._api import WeightsEnum, Weights
from ._meta import _IMAGENET_CATEGORIES
from ._utils import handle_legacy_interface, _ovewrite_named_param, _make_divisible


__all__ = [
    "MobileNetV3",
    "MobileNet_V3_Large_Weights",
    "MobileNet_V3_Small_Weights",
    "mobilenet_v3_large",
    "mobilenet_v3_small",
]


class SqueezeExcitation(SElayer):
    """DEPRECATED"""

    def __init__(self, input_channels: int, squeeze_factor: int = 4):
        squeeze_channels = _make_divisible(input_channels // squeeze_factor, 8)
        super().__init__(input_channels, squeeze_channels, scale_activation=nn.Hardsigmoid)
        self.relu = self.activation
        delattr(self, "activation")
        warnings.warn(
            "This SqueezeExcitation class is deprecated since 0.12 and will be removed in 0.14. "
            "Use torchvision.ops.SqueezeExcitation instead.",
            FutureWarning,
        )


class InvertedResidualConfig:
    # Stores information listed at Tables 1 and 2 of the MobileNetV3 paper
    def __init__(
        self,
        input_channels: int,
        kernel: int,
        expanded_channels: int,
        out_channels: int,
        use_se: bool,
        activation: str,
        stride: int,
        dilation: int,
        width_mult: float,
    ):
        self.input_channels = self.adjust_channels(input_channels, width_mult)
        self.kernel = kernel
        self.expanded_channels = self.adjust_channels(expanded_channels, width_mult)
        self.out_channels = self.adjust_channels(out_channels, width_mult)
        self.use_se = use_se
        self.use_hs = activation == "HS"
        self.stride = stride
        self.dilation = dilation

    @staticmethod
    def adjust_channels(channels: int, width_mult: float):
        return _make_divisible(channels * width_mult, 8)


class InvertedResidual(nn.Module):
    # Implemented as described at section 5 of MobileNetV3 paper
    def __init__(
        self,
        cnf: InvertedResidualConfig,
        norm_layer: Callable[..., nn.Module],
        se_layer: Callable[..., nn.Module] = partial(SElayer, scale_activation=nn.Hardsigmoid),
    ):
        super().__init__()
        if not (1 <= cnf.stride <= 2):
            raise ValueError("illegal stride value")

        self.use_res_connect = cnf.stride == 1 and cnf.input_channels == cnf.out_channels

        layers: List[nn.Module] = []
        activation_layer = nn.Hardswish if cnf.use_hs else nn.ReLU

        # expand
        if cnf.expanded_channels != cnf.input_channels:
            layers.append(
                Conv2dNormActivation(
                    cnf.input_channels,
                    cnf.expanded_channels,
                    kernel_size=1,
                    norm_layer=norm_layer,
                    activation_layer=activation_layer,
                )
            )

        # depthwise
        stride = 1 if cnf.dilation > 1 else cnf.stride
        layers.append(
            Conv2dNormActivation(
                cnf.expanded_channels,
                cnf.expanded_channels,
                kernel_size=cnf.kernel,
                stride=stride,
                dilation=cnf.dilation,
                groups=cnf.expanded_channels,
                norm_layer=norm_layer,
                activation_layer=activation_layer,
            )
        )
        if cnf.use_se:
            squeeze_channels = _make_divisible(cnf.expanded_channels // 4, 8)
            layers.append(se_layer(cnf.expanded_channels, squeeze_channels))

        # project
        layers.append(
            Conv2dNormActivation(
                cnf.expanded_channels, cnf.out_channels, kernel_size=1, norm_layer=norm_layer, activation_layer=None
            )
        )

        self.block = nn.Sequential(*layers)
        self.out_channels = cnf.out_channels
        self._is_cn = cnf.stride > 1

    def forward(self, input: Tensor) -> Tensor:
        result = self.block(input)
        if self.use_res_connect:
            result += input
        return result


class MobileNetV3(nn.Module):
    def __init__(
        self,
        inverted_residual_setting: List[InvertedResidualConfig],
        last_channel: int,
        num_classes: int = 1000,
        block: Optional[Callable[..., nn.Module]] = None,
        norm_layer: Optional[Callable[..., nn.Module]] = None,
        dropout: float = 0.2,
        **kwargs: Any,
    ) -> None:
        """
        MobileNet V3 main class

        Args:
            inverted_residual_setting (List[InvertedResidualConfig]): Network structure
            last_channel (int): The number of channels on the penultimate layer
            num_classes (int): Number of classes
            block (Optional[Callable[..., nn.Module]]): Module specifying inverted residual building block for mobilenet
            norm_layer (Optional[Callable[..., nn.Module]]): Module specifying the normalization layer to use
            dropout (float): The droupout probability
        """
        super().__init__()
        _log_api_usage_once(self)

        if not inverted_residual_setting:
            raise ValueError("The inverted_residual_setting should not be empty")
        elif not (
            isinstance(inverted_residual_setting, Sequence)
            and all([isinstance(s, InvertedResidualConfig) for s in inverted_residual_setting])
        ):
            raise TypeError("The inverted_residual_setting should be List[InvertedResidualConfig]")

        if block is None:
            block = InvertedResidual

        if norm_layer is None:
            norm_layer = partial(nn.BatchNorm2d, eps=0.001, momentum=0.01)

        layers: List[nn.Module] = []

        # building first layer
        firstconv_output_channels = inverted_residual_setting[0].input_channels
        layers.append(
            Conv2dNormActivation(
                3,
                firstconv_output_channels,
                kernel_size=3,
                stride=2,
                norm_layer=norm_layer,
                activation_layer=nn.Hardswish,
            )
        )

        # building inverted residual blocks
        for cnf in inverted_residual_setting:
            layers.append(block(cnf, norm_layer))

        # building last several layers
        lastconv_input_channels = inverted_residual_setting[-1].out_channels
        lastconv_output_channels = 6 * lastconv_input_channels
        layers.append(
            Conv2dNormActivation(
                lastconv_input_channels,
                lastconv_output_channels,
                kernel_size=1,
                norm_layer=norm_layer,
                activation_layer=nn.Hardswish,
            )
        )

        self.features = nn.Sequential(*layers)
        self.avgpool = nn.AdaptiveAvgPool2d(1)
        self.classifier = nn.Sequential(
            nn.Linear(lastconv_output_channels, last_channel),
            nn.Hardswish(inplace=True),
            nn.Dropout(p=dropout, inplace=True),
            nn.Linear(last_channel, num_classes),
        )

        for m in self.modules():
            if isinstance(m, nn.Conv2d):
                nn.init.kaiming_normal_(m.weight, mode="fan_out")
                if m.bias is not None:
                    nn.init.zeros_(m.bias)
            elif isinstance(m, (nn.BatchNorm2d, nn.GroupNorm)):
                nn.init.ones_(m.weight)
                nn.init.zeros_(m.bias)
            elif isinstance(m, nn.Linear):
                nn.init.normal_(m.weight, 0, 0.01)
                nn.init.zeros_(m.bias)

    def _forward_impl(self, x: Tensor) -> Tensor:
        x = self.features(x)

        x = self.avgpool(x)
        x = torch.flatten(x, 1)

        x = self.classifier(x)

        return x

    def forward(self, x: Tensor) -> Tensor:
        return self._forward_impl(x)


def _mobilenet_v3_conf(
    arch: str, width_mult: float = 1.0, reduced_tail: bool = False, dilated: bool = False, **kwargs: Any
):
    reduce_divider = 2 if reduced_tail else 1
    dilation = 2 if dilated else 1

    bneck_conf = partial(InvertedResidualConfig, width_mult=width_mult)
    adjust_channels = partial(InvertedResidualConfig.adjust_channels, width_mult=width_mult)

    if arch == "mobilenet_v3_large":
        inverted_residual_setting = [
            bneck_conf(16, 3, 16, 16, False, "RE", 1, 1),
            bneck_conf(16, 3, 64, 24, False, "RE", 2, 1),  # C1
            bneck_conf(24, 3, 72, 24, False, "RE", 1, 1),
            bneck_conf(24, 5, 72, 40, True, "RE", 2, 1),  # C2
            bneck_conf(40, 5, 120, 40, True, "RE", 1, 1),
            bneck_conf(40, 5, 120, 40, True, "RE", 1, 1),
            bneck_conf(40, 3, 240, 80, False, "HS", 2, 1),  # C3
            bneck_conf(80, 3, 200, 80, False, "HS", 1, 1),
            bneck_conf(80, 3, 184, 80, False, "HS", 1, 1),
            bneck_conf(80, 3, 184, 80, False, "HS", 1, 1),
            bneck_conf(80, 3, 480, 112, True, "HS", 1, 1),
            bneck_conf(112, 3, 672, 112, True, "HS", 1, 1),
            bneck_conf(112, 5, 672, 160 // reduce_divider, True, "HS", 2, dilation),  # C4
            bneck_conf(160 // reduce_divider, 5, 960 // reduce_divider, 160 // reduce_divider, True, "HS", 1, dilation),
            bneck_conf(160 // reduce_divider, 5, 960 // reduce_divider, 160 // reduce_divider, True, "HS", 1, dilation),
        ]
        last_channel = adjust_channels(1280 // reduce_divider)  # C5
    elif arch == "mobilenet_v3_small":
        inverted_residual_setting = [
            bneck_conf(16, 3, 16, 16, True, "RE", 2, 1),  # C1
            bneck_conf(16, 3, 72, 24, False, "RE", 2, 1),  # C2
            bneck_conf(24, 3, 88, 24, False, "RE", 1, 1),
            bneck_conf(24, 5, 96, 40, True, "HS", 2, 1),  # C3
            bneck_conf(40, 5, 240, 40, True, "HS", 1, 1),
            bneck_conf(40, 5, 240, 40, True, "HS", 1, 1),
            bneck_conf(40, 5, 120, 48, True, "HS", 1, 1),
            bneck_conf(48, 5, 144, 48, True, "HS", 1, 1),
            bneck_conf(48, 5, 288, 96 // reduce_divider, True, "HS", 2, dilation),  # C4
            bneck_conf(96 // reduce_divider, 5, 576 // reduce_divider, 96 // reduce_divider, True, "HS", 1, dilation),
            bneck_conf(96 // reduce_divider, 5, 576 // reduce_divider, 96 // reduce_divider, True, "HS", 1, dilation),
        ]
        last_channel = adjust_channels(1024 // reduce_divider)  # C5
    else:
        raise ValueError(f"Unsupported model type {arch}")

    return inverted_residual_setting, last_channel


def _mobilenet_v3(
    inverted_residual_setting: List[InvertedResidualConfig],
    last_channel: int,
    weights: Optional[WeightsEnum],
    progress: bool,
    **kwargs: Any,
) -> MobileNetV3:
    if weights is not None:
        _ovewrite_named_param(kwargs, "num_classes", len(weights.meta["categories"]))

    model = MobileNetV3(inverted_residual_setting, last_channel, **kwargs)

    if weights is not None:
        model.load_state_dict(weights.get_state_dict(progress=progress))

    return model


_COMMON_META = {
    "min_size": (1, 1),
    "categories": _IMAGENET_CATEGORIES,
}


class MobileNet_V3_Large_Weights(WeightsEnum):
    IMAGENET1K_V1 = Weights(
        url="https://download.pytorch.org/models/mobilenet_v3_large-8738ca79.pth",
        transforms=partial(ImageClassification, crop_size=224),
        meta={
            **_COMMON_META,
            "num_params": 5483032,
            "recipe": "https://github.com/pytorch/vision/tree/main/references/classification#mobilenetv3-large--small",
            "metrics": {
                "acc@1": 74.042,
                "acc@5": 91.340,
            },
        },
    )
    IMAGENET1K_V2 = Weights(
        url="https://download.pytorch.org/models/mobilenet_v3_large-5c1a4163.pth",
        transforms=partial(ImageClassification, crop_size=224, resize_size=232),
        meta={
            **_COMMON_META,
            "num_params": 5483032,
            "recipe": "https://github.com/pytorch/vision/issues/3995#new-recipe-with-reg-tuning",
            "metrics": {
                "acc@1": 75.274,
                "acc@5": 92.566,
            },
        },
    )
    DEFAULT = IMAGENET1K_V2


class MobileNet_V3_Small_Weights(WeightsEnum):
    IMAGENET1K_V1 = Weights(
        url="https://download.pytorch.org/models/mobilenet_v3_small-047dcff4.pth",
        transforms=partial(ImageClassification, crop_size=224),
        meta={
            **_COMMON_META,
            "num_params": 2542856,
            "recipe": "https://github.com/pytorch/vision/tree/main/references/classification#mobilenetv3-large--small",
            "metrics": {
                "acc@1": 67.668,
                "acc@5": 87.402,
            },
        },
    )
    DEFAULT = IMAGENET1K_V1


@handle_legacy_interface(weights=("pretrained", MobileNet_V3_Large_Weights.IMAGENET1K_V1))
def mobilenet_v3_large(
    *, weights: Optional[MobileNet_V3_Large_Weights] = None, progress: bool = True, **kwargs: Any
) -> MobileNetV3:
    """
    Constructs a large MobileNetV3 architecture from
    `Searching for MobileNetV3 <https://arxiv.org/abs/1905.02244>`__.

    Args:
        weights (:class:`~torchvision.models.MobileNet_V3_Large_Weights`, optional): The
            pretrained weights to use. See
            :class:`~torchvision.models.MobileNet_V3_Large_Weights` below for
            more details, and possible values. By default, no pre-trained
            weights are used.
        progress (bool, optional): If True, displays a progress bar of the
            download to stderr. Default is True.
        **kwargs: parameters passed to the ``torchvision.models.resnet.MobileNetV3``
            base class. Please refer to the `source code
            <https://github.com/pytorch/vision/blob/main/torchvision/models/mobilenetv3.py>`_
            for more details about this class.

    .. autoclass:: torchvision.models.MobileNet_V3_Large_Weights
        :members:
    """
    weights = MobileNet_V3_Large_Weights.verify(weights)

    inverted_residual_setting, last_channel = _mobilenet_v3_conf("mobilenet_v3_large", **kwargs)
    return _mobilenet_v3(inverted_residual_setting, last_channel, weights, progress, **kwargs)


@handle_legacy_interface(weights=("pretrained", MobileNet_V3_Small_Weights.IMAGENET1K_V1))
def mobilenet_v3_small(
    *, weights: Optional[MobileNet_V3_Small_Weights] = None, progress: bool = True, **kwargs: Any
) -> MobileNetV3:
    """
    Constructs a small MobileNetV3 architecture from
    `Searching for MobileNetV3 <https://arxiv.org/abs/1905.02244>`__.

    Args:
        weights (:class:`~torchvision.models.MobileNet_V3_Small_Weights`, optional): The
            pretrained weights to use. See
            :class:`~torchvision.models.MobileNet_V3_Small_Weights` below for
            more details, and possible values. By default, no pre-trained
            weights are used.
        progress (bool, optional): If True, displays a progress bar of the
            download to stderr. Default is True.
        **kwargs: parameters passed to the ``torchvision.models.resnet.MobileNetV3``
            base class. Please refer to the `source code
            <https://github.com/pytorch/vision/blob/main/torchvision/models/mobilenetv3.py>`_
            for more details about this class.

    .. autoclass:: torchvision.models.MobileNet_V3_Small_Weights
        :members:
    """
    weights = MobileNet_V3_Small_Weights.verify(weights)

    inverted_residual_setting, last_channel = _mobilenet_v3_conf("mobilenet_v3_small", **kwargs)
    return _mobilenet_v3(inverted_residual_setting, last_channel, weights, progress, **kwargs)


# The dictionary below is internal implementation detail and will be removed in v0.15
from ._utils import _ModelURLs


model_urls = _ModelURLs(
    {
        "mobilenet_v3_large": MobileNet_V3_Large_Weights.IMAGENET1K_V1.url,
        "mobilenet_v3_small": MobileNet_V3_Small_Weights.IMAGENET1K_V1.url,
    }
)