custom_resnet.py

"""
This file is a modification of the original ResNet implementation from:
https://github.com/pytorch/vision/blob/bf01bab6125c5f1152e4f336b470399e52a8559d/torchvision/models/resnet.py
"""

from functools import partial
from typing import Any, Callable, List, Optional, Type, Union

import torch
import torch.nn as nn
from torch import Tensor
from torchvision.models._api import Weights, WeightsEnum, register_model
from torchvision.models._meta import _IMAGENET_CATEGORIES
from torchvision.models._utils import _ovewrite_named_param, handle_legacy_interface
from torchvision.transforms._presets import ImageClassification
from torchvision.utils import _log_api_usage_once

__all__ = [
    "ResNet",
    "ResNet18_Weights",
    "ResNet34_Weights",
    "ResNet50_Weights",
    "ResNet101_Weights",
    "ResNet152_Weights",
    "ResNeXt50_32X4D_Weights",
    "ResNeXt101_32X8D_Weights",
    "ResNeXt101_64X4D_Weights",
    "Wide_ResNet50_2_Weights",
    "Wide_ResNet101_2_Weights",
    "resnet18",
    "resnet34",
    "resnet50",
    "resnet101",
    "resnet152",
    "resnext50_32x4d",
    "resnext101_32x8d",
    "resnext101_64x4d",
    "wide_resnet50_2",
    "wide_resnet101_2",
]


def conv3x3(
    in_planes: int, out_planes: int, stride: int = 1, groups: int = 1, dilation: int = 1
) -> nn.Conv2d:
    """3x3 convolution with padding"""
    return nn.Conv2d(
        in_planes,
        out_planes,
        kernel_size=3,
        stride=stride,
        padding=dilation,
        groups=groups,
        bias=False,
        dilation=dilation,
    )


def conv1x1(in_planes: int, out_planes: int, stride: int = 1) -> nn.Conv2d:
    """1x1 convolution"""
    return nn.Conv2d(in_planes, out_planes, kernel_size=1, stride=stride, bias=False)


class BasicBlock(nn.Module):
    expansion: int = 1

    def __init__(
        self,
        inplanes: int,
        planes: int,
        stride: int = 1,
        downsample: Optional[nn.Module] = None,
        groups: int = 1,
        base_width: int = 64,
        dilation: int = 1,
        norm_layer: Optional[Callable[..., nn.Module]] = None,
    ) -> None:
        super().__init__()
        if norm_layer is None:
            norm_layer = nn.BatchNorm2d
        if groups != 1 or base_width != 64:
            raise ValueError("BasicBlock only supports groups=1 and base_width=64")
        if dilation > 1:
            raise NotImplementedError("Dilation > 1 not supported in BasicBlock")
        # Both self.conv1 and self.downsample layers downsample the input when stride != 1
        self.conv1 = conv3x3(inplanes, planes, stride)
        self.bn1 = norm_layer(planes)
        self.relu = nn.ReLU(inplace=True)
        self.conv2 = conv3x3(planes, planes)
        self.bn2 = norm_layer(planes)
        self.downsample = downsample
        self.stride = stride

    def forward(self, x: Tensor) -> Tensor:
        identity = x

        out = self.conv1(x)
        out = self.bn1(out)
        out = self.relu(out)

        out = self.conv2(out)
        out = self.bn2(out)

        if self.downsample is not None:
            identity = self.downsample(x)

        out += identity
        out = self.relu(out)

        return out


class Bottleneck(nn.Module):
    # Bottleneck in torchvision places the stride for downsampling at 3x3 convolution(self.conv2)
    # while original implementation places the stride at the first 1x1 convolution(self.conv1)
    # according to "Deep residual learning for image recognition" https://arxiv.org/abs/1512.03385.
    # This variant is also known as ResNet V1.5 and improves accuracy according to
    # https://ngc.nvidia.com/catalog/model-scripts/nvidia:resnet_50_v1_5_for_pytorch.

    expansion: int = 4

    def __init__(
        self,
        inplanes: int,
        planes: int,
        stride: int = 1,
        downsample: Optional[nn.Module] = None,
        groups: int = 1,
        base_width: int = 64,
        dilation: int = 1,
        norm_layer: Optional[Callable[..., nn.Module]] = None,
    ) -> None:
        super().__init__()
        if norm_layer is None:
            norm_layer = nn.BatchNorm2d
        width = int(planes * (base_width / 64.0)) * groups
        # Both self.conv2 and self.downsample layers downsample the input when stride != 1
        self.conv1 = conv1x1(inplanes, width)
        self.bn1 = norm_layer(width)
        self.conv2 = conv3x3(width, width, stride, groups, dilation)
        self.bn2 = norm_layer(width)
        self.conv3 = conv1x1(width, planes * self.expansion)
        self.bn3 = norm_layer(planes * self.expansion)
        self.relu = nn.ReLU(inplace=True)
        self.downsample = downsample
        self.stride = stride

    def forward(self, x: Tensor) -> Tensor:
        identity = x

        out = self.conv1(x)
        out = self.bn1(out)
        out = self.relu(out)

        out = self.conv2(out)
        out = self.bn2(out)
        out = self.relu(out)

        out = self.conv3(out)
        out = self.bn3(out)

        if self.downsample is not None:
            identity = self.downsample(x)

        out += identity
        out = self.relu(out)

        return out


class ResNet(nn.Module):
    def __init__(
        self,
        block: Type[Union[BasicBlock, Bottleneck]],
        layers: List[int],
        num_classes: int = 1000,
        zero_init_residual: bool = False,
        groups: int = 1,
        width_per_group: int = 64,
        replace_stride_with_dilation: Optional[List[bool]] = None,
        norm_layer: Optional[Callable[..., nn.Module]] = None,
    ) -> None:
        super().__init__()
        _log_api_usage_once(self)
        if norm_layer is None:
            norm_layer = nn.BatchNorm2d
        self._norm_layer = norm_layer

        self.inplanes = 64
        self.dilation = 1
        if replace_stride_with_dilation is None:
            # each element in the tuple indicates if we should replace
            # the 2x2 stride with a dilated convolution instead
            replace_stride_with_dilation = [False, False, False]
        if len(replace_stride_with_dilation) != 3:
            raise ValueError(
                "replace_stride_with_dilation should be None "
                f"or a 3-element tuple, got {replace_stride_with_dilation}"
            )
        self.groups = groups
        self.base_width = width_per_group
        self.conv1 = nn.Conv2d(3, self.inplanes, kernel_size=7, stride=2, padding=3, bias=False)
        self.bn1 = norm_layer(self.inplanes)
        self.relu = nn.ReLU(inplace=True)
        self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1)
        self.layer1 = self._make_layer(block, 64, layers[0])
        self.layer2 = self._make_layer(
            block, 128, layers[1], stride=2, dilate=replace_stride_with_dilation[0]
        )
        self.layer3 = self._make_layer(
            block, 256, layers[2], stride=2, dilate=replace_stride_with_dilation[1]
        )
        self.layer4 = self._make_layer(
            block, 512, layers[3], stride=2, dilate=replace_stride_with_dilation[2]
        )
        # self.avgpool = nn.AdaptiveAvgPool2d((1, 1))  # FIXME
        self.avgpool = nn.AvgPool2d(kernel_size=7, stride=1, padding=0)
        self.fc = nn.Linear(512 * block.expansion, num_classes)

        for m in self.modules():
            if isinstance(m, nn.Conv2d):
                nn.init.kaiming_normal_(m.weight, mode="fan_out", nonlinearity="relu")
            elif isinstance(m, (nn.BatchNorm2d, nn.GroupNorm)):
                nn.init.constant_(m.weight, 1)
                nn.init.constant_(m.bias, 0)

        # Zero-initialize the last BN in each residual branch,
        # so that the residual branch starts with zeros, and each residual block behaves like an identity.
        # This improves the model by 0.2~0.3% according to https://arxiv.org/abs/1706.02677
        if zero_init_residual:
            for m in self.modules():
                if isinstance(m, Bottleneck) and m.bn3.weight is not None:
                    nn.init.constant_(m.bn3.weight, 0)  # type: ignore[arg-type]
                elif isinstance(m, BasicBlock) and m.bn2.weight is not None:
                    nn.init.constant_(m.bn2.weight, 0)  # type: ignore[arg-type]

    def _make_layer(
        self,
        block: Type[Union[BasicBlock, Bottleneck]],
        planes: int,
        blocks: int,
        stride: int = 1,
        dilate: bool = False,
    ) -> nn.Sequential:
        norm_layer = self._norm_layer
        downsample = None
        previous_dilation = self.dilation
        if dilate:
            self.dilation *= stride
            stride = 1
        if stride != 1 or self.inplanes != planes * block.expansion:
            downsample = nn.Sequential(
                conv1x1(self.inplanes, planes * block.expansion, stride),
                norm_layer(planes * block.expansion),
            )

        layers = []
        layers.append(
            block(
                self.inplanes,
                planes,
                stride,
                downsample,
                self.groups,
                self.base_width,
                previous_dilation,
                norm_layer,
            )
        )
        self.inplanes = planes * block.expansion
        for _ in range(1, blocks):
            layers.append(
                block(
                    self.inplanes,
                    planes,
                    groups=self.groups,
                    base_width=self.base_width,
                    dilation=self.dilation,
                    norm_layer=norm_layer,
                )
            )

        return nn.Sequential(*layers)

    def _forward_impl(self, x: Tensor) -> Tensor:
        # See note [TorchScript super()]
        x = self.conv1(x)
        x = self.bn1(x)
        x = self.relu(x)
        x = self.maxpool(x)

        x = self.layer1(x)
        x = self.layer2(x)
        x = self.layer3(x)
        x = self.layer4(x)

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

        return x

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


def _resnet(
    block: Type[Union[BasicBlock, Bottleneck]],
    layers: List[int],
    weights: Optional[WeightsEnum],
    progress: bool,
    **kwargs: Any,
) -> ResNet:
    if weights is not None:
        _ovewrite_named_param(kwargs, "num_classes", len(weights.meta["categories"]))

    model = ResNet(block, layers, **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 ResNet18_Weights(WeightsEnum):
    IMAGENET1K_V1 = Weights(
        url="https://download.pytorch.org/models/resnet18-f37072fd.pth",
        transforms=partial(ImageClassification, crop_size=224),
        meta={
            **_COMMON_META,
            "num_params": 11689512,
            "recipe": "https://github.com/pytorch/vision/tree/main/references/classification#resnet",
            "_metrics": {
                "ImageNet-1K": {
                    "acc@1": 69.758,
                    "acc@5": 89.078,
                }
            },
            "_ops": 1.814,
            "_file_size": 44.661,
            "_docs": """These weights reproduce closely the results of the paper using a simple training recipe.""",
        },
    )
    DEFAULT = IMAGENET1K_V1


class ResNet34_Weights(WeightsEnum):
    IMAGENET1K_V1 = Weights(
        url="https://download.pytorch.org/models/resnet34-b627a593.pth",
        transforms=partial(ImageClassification, crop_size=224),
        meta={
            **_COMMON_META,
            "num_params": 21797672,
            "recipe": "https://github.com/pytorch/vision/tree/main/references/classification#resnet",
            "_metrics": {
                "ImageNet-1K": {
                    "acc@1": 73.314,
                    "acc@5": 91.420,
                }
            },
            "_ops": 3.664,
            "_file_size": 83.275,
            "_docs": """These weights reproduce closely the results of the paper using a simple training recipe.""",
        },
    )
    DEFAULT = IMAGENET1K_V1


class ResNet50_Weights(WeightsEnum):
    IMAGENET1K_V1 = Weights(
        url="https://download.pytorch.org/models/resnet50-0676ba61.pth",
        transforms=partial(ImageClassification, crop_size=224),
        meta={
            **_COMMON_META,
            "num_params": 25557032,
            "recipe": "https://github.com/pytorch/vision/tree/main/references/classification#resnet",
            "_metrics": {
                "ImageNet-1K": {
                    "acc@1": 76.130,
                    "acc@5": 92.862,
                }
            },
            "_ops": 4.089,
            "_file_size": 97.781,
            "_docs": """These weights reproduce closely the results of the paper using a simple training recipe.""",
        },
    )
    IMAGENET1K_V2 = Weights(
        url="https://download.pytorch.org/models/resnet50-11ad3fa6.pth",
        transforms=partial(ImageClassification, crop_size=224, resize_size=232),
        meta={
            **_COMMON_META,
            "num_params": 25557032,
            "recipe": "https://github.com/pytorch/vision/issues/3995#issuecomment-1013906621",
            "_metrics": {
                "ImageNet-1K": {
                    "acc@1": 80.858,
                    "acc@5": 95.434,
                }
            },
            "_ops": 4.089,
            "_file_size": 97.79,
            "_docs": """
                These weights improve upon the results of the original paper by using TorchVision's `new training recipe
                <https://pytorch.org/blog/how-to-train-state-of-the-art-models-using-torchvision-latest-primitives/>`_.
            """,
        },
    )
    DEFAULT = IMAGENET1K_V2


class ResNet101_Weights(WeightsEnum):
    IMAGENET1K_V1 = Weights(
        url="https://download.pytorch.org/models/resnet101-63fe2227.pth",
        transforms=partial(ImageClassification, crop_size=224),
        meta={
            **_COMMON_META,
            "num_params": 44549160,
            "recipe": "https://github.com/pytorch/vision/tree/main/references/classification#resnet",
            "_metrics": {
                "ImageNet-1K": {
                    "acc@1": 77.374,
                    "acc@5": 93.546,
                }
            },
            "_ops": 7.801,
            "_file_size": 170.511,
            "_docs": """These weights reproduce closely the results of the paper using a simple training recipe.""",
        },
    )
    IMAGENET1K_V2 = Weights(
        url="https://download.pytorch.org/models/resnet101-cd907fc2.pth",
        transforms=partial(ImageClassification, crop_size=224, resize_size=232),
        meta={
            **_COMMON_META,
            "num_params": 44549160,
            "recipe": "https://github.com/pytorch/vision/issues/3995#new-recipe",
            "_metrics": {
                "ImageNet-1K": {
                    "acc@1": 81.886,
                    "acc@5": 95.780,
                }
            },
            "_ops": 7.801,
            "_file_size": 170.53,
            "_docs": """
                These weights improve upon the results of the original paper by using TorchVision's `new training recipe
                <https://pytorch.org/blog/how-to-train-state-of-the-art-models-using-torchvision-latest-primitives/>`_.
            """,
        },
    )
    DEFAULT = IMAGENET1K_V2


class ResNet152_Weights(WeightsEnum):
    IMAGENET1K_V1 = Weights(
        url="https://download.pytorch.org/models/resnet152-394f9c45.pth",
        transforms=partial(ImageClassification, crop_size=224),
        meta={
            **_COMMON_META,
            "num_params": 60192808,
            "recipe": "https://github.com/pytorch/vision/tree/main/references/classification#resnet",
            "_metrics": {
                "ImageNet-1K": {
                    "acc@1": 78.312,
                    "acc@5": 94.046,
                }
            },
            "_ops": 11.514,
            "_file_size": 230.434,
            "_docs": """These weights reproduce closely the results of the paper using a simple training recipe.""",
        },
    )
    IMAGENET1K_V2 = Weights(
        url="https://download.pytorch.org/models/resnet152-f82ba261.pth",
        transforms=partial(ImageClassification, crop_size=224, resize_size=232),
        meta={
            **_COMMON_META,
            "num_params": 60192808,
            "recipe": "https://github.com/pytorch/vision/issues/3995#new-recipe",
            "_metrics": {
                "ImageNet-1K": {
                    "acc@1": 82.284,
                    "acc@5": 96.002,
                }
            },
            "_ops": 11.514,
            "_file_size": 230.474,
            "_docs": """
                These weights improve upon the results of the original paper by using TorchVision's `new training recipe
                <https://pytorch.org/blog/how-to-train-state-of-the-art-models-using-torchvision-latest-primitives/>`_.
            """,
        },
    )
    DEFAULT = IMAGENET1K_V2


class ResNeXt50_32X4D_Weights(WeightsEnum):
    IMAGENET1K_V1 = Weights(
        url="https://download.pytorch.org/models/resnext50_32x4d-7cdf4587.pth",
        transforms=partial(ImageClassification, crop_size=224),
        meta={
            **_COMMON_META,
            "num_params": 25028904,
            "recipe": "https://github.com/pytorch/vision/tree/main/references/classification#resnext",
            "_metrics": {
                "ImageNet-1K": {
                    "acc@1": 77.618,
                    "acc@5": 93.698,
                }
            },
            "_ops": 4.23,
            "_file_size": 95.789,
            "_docs": """These weights reproduce closely the results of the paper using a simple training recipe.""",
        },
    )
    IMAGENET1K_V2 = Weights(
        url="https://download.pytorch.org/models/resnext50_32x4d-1a0047aa.pth",
        transforms=partial(ImageClassification, crop_size=224, resize_size=232),
        meta={
            **_COMMON_META,
            "num_params": 25028904,
            "recipe": "https://github.com/pytorch/vision/issues/3995#new-recipe",
            "_metrics": {
                "ImageNet-1K": {
                    "acc@1": 81.198,
                    "acc@5": 95.340,
                }
            },
            "_ops": 4.23,
            "_file_size": 95.833,
            "_docs": """
                These weights improve upon the results of the original paper by using TorchVision's `new training recipe
                <https://pytorch.org/blog/how-to-train-state-of-the-art-models-using-torchvision-latest-primitives/>`_.
            """,
        },
    )
    DEFAULT = IMAGENET1K_V2


class ResNeXt101_32X8D_Weights(WeightsEnum):
    IMAGENET1K_V1 = Weights(
        url="https://download.pytorch.org/models/resnext101_32x8d-8ba56ff5.pth",
        transforms=partial(ImageClassification, crop_size=224),
        meta={
            **_COMMON_META,
            "num_params": 88791336,
            "recipe": "https://github.com/pytorch/vision/tree/main/references/classification#resnext",
            "_metrics": {
                "ImageNet-1K": {
                    "acc@1": 79.312,
                    "acc@5": 94.526,
                }
            },
            "_ops": 16.414,
            "_file_size": 339.586,
            "_docs": """These weights reproduce closely the results of the paper using a simple training recipe.""",
        },
    )
    IMAGENET1K_V2 = Weights(
        url="https://download.pytorch.org/models/resnext101_32x8d-110c445d.pth",
        transforms=partial(ImageClassification, crop_size=224, resize_size=232),
        meta={
            **_COMMON_META,
            "num_params": 88791336,
            "recipe": "https://github.com/pytorch/vision/issues/3995#new-recipe-with-fixres",
            "_metrics": {
                "ImageNet-1K": {
                    "acc@1": 82.834,
                    "acc@5": 96.228,
                }
            },
            "_ops": 16.414,
            "_file_size": 339.673,
            "_docs": """
                These weights improve upon the results of the original paper by using TorchVision's `new training recipe
                <https://pytorch.org/blog/how-to-train-state-of-the-art-models-using-torchvision-latest-primitives/>`_.
            """,
        },
    )
    DEFAULT = IMAGENET1K_V2


class ResNeXt101_64X4D_Weights(WeightsEnum):
    IMAGENET1K_V1 = Weights(
        url="https://download.pytorch.org/models/resnext101_64x4d-173b62eb.pth",
        transforms=partial(ImageClassification, crop_size=224, resize_size=232),
        meta={
            **_COMMON_META,
            "num_params": 83455272,
            "recipe": "https://github.com/pytorch/vision/pull/5935",
            "_metrics": {
                "ImageNet-1K": {
                    "acc@1": 83.246,
                    "acc@5": 96.454,
                }
            },
            "_ops": 15.46,
            "_file_size": 319.318,
            "_docs": """
                These weights were trained from scratch by using TorchVision's `new training recipe
                <https://pytorch.org/blog/how-to-train-state-of-the-art-models-using-torchvision-latest-primitives/>`_.
            """,
        },
    )
    DEFAULT = IMAGENET1K_V1


class Wide_ResNet50_2_Weights(WeightsEnum):
    IMAGENET1K_V1 = Weights(
        url="https://download.pytorch.org/models/wide_resnet50_2-95faca4d.pth",
        transforms=partial(ImageClassification, crop_size=224),
        meta={
            **_COMMON_META,
            "num_params": 68883240,
            "recipe": "https://github.com/pytorch/vision/pull/912#issue-445437439",
            "_metrics": {
                "ImageNet-1K": {
                    "acc@1": 78.468,
                    "acc@5": 94.086,
                }
            },
            "_ops": 11.398,
            "_file_size": 131.82,
            "_docs": """These weights reproduce closely the results of the paper using a simple training recipe.""",
        },
    )
    IMAGENET1K_V2 = Weights(
        url="https://download.pytorch.org/models/wide_resnet50_2-9ba9bcbe.pth",
        transforms=partial(ImageClassification, crop_size=224, resize_size=232),
        meta={
            **_COMMON_META,
            "num_params": 68883240,
            "recipe": "https://github.com/pytorch/vision/issues/3995#new-recipe-with-fixres",
            "_metrics": {
                "ImageNet-1K": {
                    "acc@1": 81.602,
                    "acc@5": 95.758,
                }
            },
            "_ops": 11.398,
            "_file_size": 263.124,
            "_docs": """
                These weights improve upon the results of the original paper by using TorchVision's `new training recipe
                <https://pytorch.org/blog/how-to-train-state-of-the-art-models-using-torchvision-latest-primitives/>`_.
            """,
        },
    )
    DEFAULT = IMAGENET1K_V2


class Wide_ResNet101_2_Weights(WeightsEnum):
    IMAGENET1K_V1 = Weights(
        url="https://download.pytorch.org/models/wide_resnet101_2-32ee1156.pth",
        transforms=partial(ImageClassification, crop_size=224),
        meta={
            **_COMMON_META,
            "num_params": 126886696,
            "recipe": "https://github.com/pytorch/vision/pull/912#issue-445437439",
            "_metrics": {
                "ImageNet-1K": {
                    "acc@1": 78.848,
                    "acc@5": 94.284,
                }
            },
            "_ops": 22.753,
            "_file_size": 242.896,
            "_docs": """These weights reproduce closely the results of the paper using a simple training recipe.""",
        },
    )
    IMAGENET1K_V2 = Weights(
        url="https://download.pytorch.org/models/wide_resnet101_2-d733dc28.pth",
        transforms=partial(ImageClassification, crop_size=224, resize_size=232),
        meta={
            **_COMMON_META,
            "num_params": 126886696,
            "recipe": "https://github.com/pytorch/vision/issues/3995#new-recipe",
            "_metrics": {
                "ImageNet-1K": {
                    "acc@1": 82.510,
                    "acc@5": 96.020,
                }
            },
            "_ops": 22.753,
            "_file_size": 484.747,
            "_docs": """
                These weights improve upon the results of the original paper by using TorchVision's `new training recipe
                <https://pytorch.org/blog/how-to-train-state-of-the-art-models-using-torchvision-latest-primitives/>`_.
            """,
        },
    )
    DEFAULT = IMAGENET1K_V2


@register_model()
@handle_legacy_interface(weights=("pretrained", ResNet18_Weights.IMAGENET1K_V1))
def resnet18_custom(
    *, weights: Optional[ResNet18_Weights] = None, progress: bool = True, **kwargs: Any
) -> ResNet:
    """ResNet-18 from `Deep Residual Learning for Image Recognition <https://arxiv.org/abs/1512.03385>`__.

    Args:
        weights (:class:`~torchvision.models.ResNet18_Weights`, optional): The
            pre-trained weights to use. See
            :class:`~torchvision.models.ResNet18_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.ResNet``
            base class. Please refer to the `source code
            <https://github.com/pytorch/vision/blob/main/torchvision/models/resnet.py>`_
            for more details about this class.

    .. autoclass:: torchvision.models.ResNet18_Weights
        :members:
    """
    weights = ResNet18_Weights.verify(weights)

    return _resnet(BasicBlock, [2, 2, 2, 2], weights, progress, **kwargs)