torchmetric_fdd.py

# Copyright The Lightning team.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from copy import deepcopy
from typing import Any, List, Optional, Sequence, Tuple, Union

import torch
from torch import Tensor
from torch.nn import Module
from torch.nn.functional import adaptive_avg_pool2d

from torchmetrics.metric import Metric
from torchmetrics.utilities.imports import (
    _MATPLOTLIB_AVAILABLE,
    _TORCH_FIDELITY_AVAILABLE,
)
from torchmetrics.utilities.plot import _AX_TYPE, _PLOT_OUT_TYPE

if not _MATPLOTLIB_AVAILABLE:
    __doctest_skip__ = ["FrechetInceptionDistance.plot"]

if _TORCH_FIDELITY_AVAILABLE:
    from torch_fidelity.feature_extractor_inceptionv3 import (
        FeatureExtractorInceptionV3 as _FeatureExtractorInceptionV3,
    )
    from torch_fidelity.helpers import vassert
    from torch_fidelity.interpolate_compat_tensorflow import (
        interpolate_bilinear_2d_like_tensorflow1x,
    )
else:

    class _FeatureExtractorInceptionV3(Module):  # type: ignore[no-redef]
        pass

    vassert = None
    interpolate_bilinear_2d_like_tensorflow1x = None

    __doctest_skip__ = ["FrechetInceptionDistance", "FrechetInceptionDistance.plot"]

from abc import ABC, abstractmethod
import torchvision.transforms as TF
import numpy as np
import torch
import torch.nn as nn
import torch.nn.functional as F

import sys


import torch.nn as nn


class Encoder(ABC, nn.Module):
    def __init__(self, *args, **kwargs):
        nn.Module.__init__(self)
        self.setup(*args, **kwargs)
        self.name = "encoder"

    @abstractmethod
    def setup(self, *args, **kwargs):
        pass

    @abstractmethod
    def transform(self, x):
        """Converts a PIL Image to an input for the model"""
        pass

    def forward(self, *args, **kwargs):
        return self.model(*args, **kwargs)


class DINOv2Encoder(Encoder):
    def setup(self, arch=None, clean_resize: bool = False):
        if arch is None:
            arch = "vitl14"

        self.arch = arch

        arch_str = f"dinov2_{self.arch}"

        self.model = torch.hub.load(
            "facebookresearch/dinov2",
            arch_str,
            verbose=True,
        )  # juwels

    def transform(self, image):

        imagenet_mean = np.array([0.485, 0.456, 0.406])
        imagenet_std = np.array([0.229, 0.224, 0.225])

        image = TF.Compose(
            [
                TF.Resize((224, 224), TF.InterpolationMode.BICUBIC),
            ]
        )(image)
        image = image.to(torch.float)

        return TF.Normalize(imagenet_mean, imagenet_std)(image)


def _compute_fid(mu1: Tensor, sigma1: Tensor, mu2: Tensor, sigma2: Tensor) -> Tensor:
    r"""Compute adjusted version of `Fid Score`_.

    The Frechet Inception Distance between two multivariate Gaussians X_x ~ N(mu_1, sigm_1)
    and X_y ~ N(mu_2, sigm_2) is d^2 = ||mu_1 - mu_2||^2 + Tr(sigm_1 + sigm_2 - 2*sqrt(sigm_1*sigm_2)).

    Args:
        mu1: mean of activations calculated on predicted (x) samples
        sigma1: covariance matrix over activations calculated on predicted (x) samples
        mu2: mean of activations calculated on target (y) samples
        sigma2: covariance matrix over activations calculated on target (y) samples

    Returns:
        Scalar value of the distance between sets.

    """
    a = (mu1 - mu2).square().sum(dim=-1)
    b = sigma1.trace() + sigma2.trace()
    c = torch.linalg.eigvals(sigma1 @ sigma2).sqrt().real.sum(dim=-1)

    return a + b - 2 * c


class FrechetDinovDistance(Metric):
    r"""Calculate Fréchet inception distance (FID_) which is used to access the quality of generated images.

    .. math::
        FID = \|\mu - \mu_w\|^2 + tr(\Sigma + \Sigma_w - 2(\Sigma \Sigma_w)^{\frac{1}{2}})

    where :math:`\mathcal{N}(\mu, \Sigma)` is the multivariate normal distribution estimated from Inception v3
    (`fid ref1`_) features calculated on real life images and :math:`\mathcal{N}(\mu_w, \Sigma_w)` is the
    multivariate normal distribution estimated from Inception v3 features calculated on generated (fake) images.
    The metric was originally proposed in `fid ref1`_.

    Using the default feature extraction (Inception v3 using the original weights from `fid ref2`_), the input is
    expected to be mini-batches of 3-channel RGB images of shape ``(3xHxW)``. If argument ``normalize``
    is ``True`` images are expected to be dtype ``float`` and have values in the ``[0,1]`` range, else if
    ``normalize`` is set to ``False`` images are expected to have dtype ``uint8`` and take values in the ``[0, 255]``
    range. All images will be resized to 299 x 299 which is the size of the original training data. The boolian
    flag ``real`` determines if the images should update the statistics of the real distribution or the
    fake distribution.

    This metric is known to be unstable in its calculatations, and we recommend for the best results using this metric
    that you calculate using `torch.float64` (default is `torch.float32`) which can be set using the `.set_dtype`
    method of the metric.

    .. note:: using this metrics requires you to have torch 1.9 or higher installed

    .. note:: using this metric with the default feature extractor requires that ``torch-fidelity``
        is installed. Either install as ``pip install torchmetrics[image]`` or ``pip install torch-fidelity``

    As input to ``forward`` and ``update`` the metric accepts the following input

    - ``imgs`` (:class:`~torch.Tensor`): tensor with images feed to the feature extractor with
    - ``real`` (:class:`~bool`): bool indicating if ``imgs`` belong to the real or the fake distribution

    As output of `forward` and `compute` the metric returns the following output

    - ``fid`` (:class:`~torch.Tensor`): float scalar tensor with mean FID value over samples

    Args:
        feature:
            Either an integer or ``nn.Module``:

            - an integer will indicate the inceptionv3 feature layer to choose. Can be one of the following:
              64, 192, 768, 2048
            - an ``nn.Module`` for using a custom feature extractor. Expects that its forward method returns
              an ``(N,d)`` matrix where ``N`` is the batch size and ``d`` is the feature size.

        reset_real_features: Whether to also reset the real features. Since in many cases the real dataset does not
            change, the features can be cached them to avoid recomputing them which is costly. Set this to ``False`` if
            your dataset does not change.
        kwargs: Additional keyword arguments, see :ref:`Metric kwargs` for more info.

    .. note::
        If a custom feature extractor is provided through the `feature` argument it is expected to either have a
        attribute called ``num_features`` that indicates the number of features returned by the forward pass or
        alternatively we will pass through tensor of shape ``(1, 3, 299, 299)`` and dtype ``torch.uint8``` to the
        forward pass and expect a tensor of shape ``(1, num_features)`` as output.

    Raises:
        ValueError:
            If torch version is lower than 1.9
        ModuleNotFoundError:
            If ``feature`` is set to an ``int`` (default settings) and ``torch-fidelity`` is not installed
        ValueError:
            If ``feature`` is set to an ``int`` not in [64, 192, 768, 2048]
        TypeError:
            If ``feature`` is not an ``str``, ``int`` or ``torch.nn.Module``
        ValueError:
            If ``reset_real_features`` is not an ``bool``

    Example:
        >>> import torch
        >>> _ = torch.manual_seed(123)
        >>> from torchmetrics.image.fid import FrechetInceptionDistance
        >>> fid = FrechetInceptionDistance(feature=64)
        >>> # generate two slightly overlapping image intensity distributions
        >>> imgs_dist1 = torch.randint(0, 200, (100, 3, 299, 299), dtype=torch.uint8)
        >>> imgs_dist2 = torch.randint(100, 255, (100, 3, 299, 299), dtype=torch.uint8)
        >>> fid.update(imgs_dist1, real=True)
        >>> fid.update(imgs_dist2, real=False)
        >>> fid.compute()
        tensor(12.7202)

    """

    higher_is_better: bool = False
    is_differentiable: bool = False
    full_state_update: bool = False
    plot_lower_bound: float = 0.0

    real_features_sum: Tensor
    real_features_cov_sum: Tensor
    real_features_num_samples: Tensor

    fake_features_sum: Tensor
    fake_features_cov_sum: Tensor
    fake_features_num_samples: Tensor

    inception: Module
    feature_network: str = "inception"

    def __init__(
        self,
        feature: Union[int, Module] = 1024,
        reset_real_features: bool = True,
        normalize: bool = False,
        **kwargs: Any,
    ) -> None:
        super().__init__(**kwargs)
        if isinstance(feature, int):
            num_features = feature
            if not _TORCH_FIDELITY_AVAILABLE:
                raise ModuleNotFoundError(
                    "FrechetInceptionDistance metric requires that `Torch-fidelity` is installed."
                    " Either install as `pip install torchmetrics[image]` or `pip install torch-fidelity`."
                )

            self.inception = DINOv2Encoder()

        elif isinstance(feature, Module):
            self.inception = feature
            if hasattr(self.inception, "num_features"):
                num_features = self.inception.num_features
            else:
                dummy_image = torch.randint(0, 255, (1, 3, 299, 299), dtype=torch.uint8)
                num_features = self.inception(dummy_image).shape[-1]
        else:
            raise TypeError("Got unknown input to argument `feature`")

        if not isinstance(reset_real_features, bool):
            raise ValueError("Argument `reset_real_features` expected to be a bool")
        self.reset_real_features = reset_real_features

        if not isinstance(normalize, bool):
            raise ValueError("Argument `normalize` expected to be a bool")
        self.normalize = normalize

        mx_num_feats = (num_features, num_features)
        self.add_state(
            "real_features_sum",
            torch.zeros(num_features).double(),
            dist_reduce_fx="sum",
        )
        self.add_state(
            "real_features_cov_sum",
            torch.zeros(mx_num_feats).double(),
            dist_reduce_fx="sum",
        )
        self.add_state(
            "real_features_num_samples", torch.tensor(0).long(), dist_reduce_fx="sum"
        )

        self.add_state(
            "fake_features_sum",
            torch.zeros(num_features).double(),
            dist_reduce_fx="sum",
        )
        self.add_state(
            "fake_features_cov_sum",
            torch.zeros(mx_num_feats).double(),
            dist_reduce_fx="sum",
        )
        self.add_state(
            "fake_features_num_samples", torch.tensor(0).long(), dist_reduce_fx="sum"
        )

    def update(self, imgs: Tensor, real: bool) -> None:
        """Update the state with extracted features."""
        imgs = (imgs * 255).byte() if self.normalize else imgs
        imgs = self.inception.transform(imgs)
        features = self.inception(imgs)
        self.orig_dtype = features.dtype
        features = features.double()

        if features.dim() == 1:
            features = features.unsqueeze(0)
        if real:
            self.real_features_sum += features.sum(dim=0)
            self.real_features_cov_sum += features.t().mm(features)
            self.real_features_num_samples += imgs.shape[0]
        else:
            self.fake_features_sum += features.sum(dim=0)
            self.fake_features_cov_sum += features.t().mm(features)
            self.fake_features_num_samples += imgs.shape[0]

    def compute(self) -> Tensor:
        """Calculate FID score based on accumulated extracted features from the two distributions."""
        if self.real_features_num_samples < 2 or self.fake_features_num_samples < 2:
            raise RuntimeError(
                "More than one sample is required for both the real and fake distributed to compute FID"
            )
        mean_real = (self.real_features_sum / self.real_features_num_samples).unsqueeze(
            0
        )
        mean_fake = (self.fake_features_sum / self.fake_features_num_samples).unsqueeze(
            0
        )

        cov_real_num = (
            self.real_features_cov_sum
            - self.real_features_num_samples * mean_real.t().mm(mean_real)
        )
        cov_real = cov_real_num / (self.real_features_num_samples - 1)
        cov_fake_num = (
            self.fake_features_cov_sum
            - self.fake_features_num_samples * mean_fake.t().mm(mean_fake)
        )
        cov_fake = cov_fake_num / (self.fake_features_num_samples - 1)
        return _compute_fid(
            mean_real.squeeze(0), cov_real, mean_fake.squeeze(0), cov_fake
        ).to(self.orig_dtype)

    def reset(self) -> None:
        """Reset metric states."""
        if not self.reset_real_features:
            real_features_sum = deepcopy(self.real_features_sum)
            real_features_cov_sum = deepcopy(self.real_features_cov_sum)
            real_features_num_samples = deepcopy(self.real_features_num_samples)
            super().reset()
            self.real_features_sum = real_features_sum
            self.real_features_cov_sum = real_features_cov_sum
            self.real_features_num_samples = real_features_num_samples
        else:
            super().reset()

    def set_dtype(self, dst_type: Union[str, torch.dtype]) -> "Metric":
        """Transfer all metric state to specific dtype. Special version of standard `type` method.

        Arguments:
            dst_type: the desired type as ``torch.dtype`` or string

        """
        out = super().set_dtype(dst_type)
        if isinstance(out.inception, NoTrainInceptionV3):
            out.inception._dtype = dst_type
        return out

    def plot(
        self,
        val: Optional[Union[Tensor, Sequence[Tensor]]] = None,
        ax: Optional[_AX_TYPE] = None,
    ) -> _PLOT_OUT_TYPE:
        """Plot a single or multiple values from the metric.

        Args:
            val: Either a single result from calling `metric.forward` or `metric.compute` or a list of these results.
                If no value is provided, will automatically call `metric.compute` and plot that result.
            ax: An matplotlib axis object. If provided will add plot to that axis

        Returns:
            Figure and Axes object

        Raises:
            ModuleNotFoundError:
                If `matplotlib` is not installed

        .. plot::
            :scale: 75

            >>> # Example plotting a single value
            >>> import torch
            >>> from torchmetrics.image.fid import FrechetInceptionDistance
            >>> imgs_dist1 = torch.randint(0, 200, (100, 3, 299, 299), dtype=torch.uint8)
            >>> imgs_dist2 = torch.randint(100, 255, (100, 3, 299, 299), dtype=torch.uint8)
            >>> metric = FrechetInceptionDistance(feature=64)
            >>> metric.update(imgs_dist1, real=True)
            >>> metric.update(imgs_dist2, real=False)
            >>> fig_, ax_ = metric.plot()

        .. plot::
            :scale: 75

            >>> # Example plotting multiple values
            >>> import torch
            >>> from torchmetrics.image.fid import FrechetInceptionDistance
            >>> imgs_dist1 = lambda: torch.randint(0, 200, (100, 3, 299, 299), dtype=torch.uint8)
            >>> imgs_dist2 = lambda: torch.randint(100, 255, (100, 3, 299, 299), dtype=torch.uint8)
            >>> metric = FrechetInceptionDistance(feature=64)
            >>> values = [ ]
            >>> for _ in range(3):
            ...     metric.update(imgs_dist1(), real=True)
            ...     metric.update(imgs_dist2(), real=False)
            ...     values.append(metric.compute())
            ...     metric.reset()
            >>> fig_, ax_ = metric.plot(values)

        """
        return self._plot(val, ax)


if __name__ == "__main__":
    import torch

    _ = torch.manual_seed(123)
    fdd = FrechetDinovDistance().to("cuda")

    imgs_dist1 = torch.randint(0, 255, (100, 3, 299, 299), dtype=torch.uint8).to("cuda")
    imgs_dist2 = torch.randint(0, 255, (100, 3, 299, 299), dtype=torch.uint8).to("cuda")
    fdd.update(imgs_dist1, real=True)
    fdd.update(imgs_dist2, real=False)
    result = fdd.compute()
    print(result)