o .wiLL@sddlmZddlmZmZmZmZddlZddlmZddl m Z ddl m Z m Z mZddlmZddlmZdd lmZdd lmZmZesLd d gZGd ddeZGdddeZdS))Sequence)AnyListOptionalUnionN)Tensor)Literal)_multiscale_ssim_update_ssim_check_inputs _ssim_update)Metric) dim_zero_cat)_MATPLOTLIB_AVAILABLE)_AX_TYPE_PLOT_OUT_TYPE%StructuralSimilarityIndexMeasure.plot/MultiScaleStructuralSimilarityIndexMeasure.plotcsJeZdZUdZdZeed<dZeed<dZeed<dZ e ed<d Z e ed <e e ed <e e ed <        d)dedee ee fdeeeefdeddeee ee e ffde de dedededdffdd Zd e d e ddfd!d"Zdee ee e fffd#d$Z d*d%eee ee fd&eedefd'd(ZZS)+ StructuralSimilarityIndexMeasurea4 Compute Structural Similarity Index Measure (SSIM_). As input to ``forward`` and ``update`` the metric accepts the following input - ``preds`` (:class:`~torch.Tensor`): Predictions from model - ``target`` (:class:`~torch.Tensor`): Ground truth values As output of `forward` and `compute` the metric returns the following output - ``ssim`` (:class:`~torch.Tensor`): if ``reduction!='none'`` returns float scalar tensor with average SSIM value over sample else returns tensor of shape ``(N,)`` with SSIM values per sample Args: preds: estimated image target: ground truth image gaussian_kernel: If ``True`` (default), a gaussian kernel is used, if ``False`` a uniform kernel is used sigma: Standard deviation of the gaussian kernel, anisotropic kernels are possible. Ignored if a uniform kernel is used kernel_size: the size of the uniform kernel, anisotropic kernels are possible. Ignored if a Gaussian kernel is used reduction: a method to reduce metric score over individual batch scores - ``'elementwise_mean'``: takes the mean - ``'sum'``: takes the sum - ``'none'`` or ``None``: no reduction will be applied data_range: the range of the data. If None, it is determined from the data (max - min). If a tuple is provided then the range is calculated as the difference and input is clamped between the values. k1: Parameter of SSIM. k2: Parameter of SSIM. return_full_image: If true, the full ``ssim`` image is returned as a second argument. Mutually exclusive with ``return_contrast_sensitivity`` return_contrast_sensitivity: If true, the constant term is returned as a second argument. The luminance term can be obtained with luminance=ssim/contrast Mutually exclusive with ``return_full_image`` kwargs: Additional keyword arguments, see :ref:`Metric kwargs` for more info. Example: >>> import torch >>> from torchmetrics.image import StructuralSimilarityIndexMeasure >>> preds = torch.rand([3, 3, 256, 256]) >>> target = preds * 0.75 >>> ssim = StructuralSimilarityIndexMeasure(data_range=1.0) >>> ssim(preds, target) tensor(0.9219) Thigher_is_betteris_differentiableFfull_state_updateplot_lower_bound?plot_upper_boundpredstarget? elementwise_meanN{Gz?Q?gaussian_kernelsigma kernel_size reductionrsumnoneN data_rangek1k2return_full_imagereturn_contrast_sensitivitykwargsreturnc  stjd i| d} || vrtd| d||dvr)|jdtdddn|jdgdd|jd tddd| s@|rH|jd gd d||_||_||_||_ ||_ ||_ ||_ ||_ | |_dS) Nr&$Argument `reduction` must be one of , but got rr' similarityrr'defaultdist_reduce_fxtotal image_returncat)super__init__ ValueError add_statetorchtensorr"r#r$r%r)r*r+r,r-) selfr"r#r$r%r)r*r+r,r-r.valid_reduction __class__r:T/home/ubuntu/sommelier/.venv/lib/python3.10/site-packages/torchmetrics/image/ssim.pyr<Zs&  z)StructuralSimilarityIndexMeasure.__init__c Cst||\}}t|||j|j|j|j|j|j|j|j }t |t r&|\}}n|}|j s.|jr>t |j t s8td|j ||jdvrnt |jtjsNtd|j|7_t |jtjsbtd|j|jd7_dSt |jt sxtd|j|dS)*Update state with predictions and targets.@Expected `self.image_return` to be a list when returning images.r2z9Expected `self.similarity` to be a Tensor for reductions.%Expected `self.total` to be a Tensor.rz>Expected `self.similarity` to be a list when reduction='none'.N)r r r"r#r$r)r*r+r,r- isinstancetupler8list TypeErrorappendr%r3r?rr'r7shape)rArrsimilarity_packr3imager:r:rEupdates<      z'StructuralSimilarityIndexMeasure.updatecCs|jdkrt|jtrt|jtr|j|j}n'td|jdkr/t|jts+td|j}nt|jtr;t|j}ntd|jsE|j rXt|j trTt|j }||fStd|S)zCompute SSIM over state.rz_Expected `self.similarity`and `self.total` to be of type Tensor for elementwise_mean reduction.r'>> # Example plotting a single value >>> import torch >>> from torchmetrics.image import StructuralSimilarityIndexMeasure >>> preds = torch.rand([3, 3, 256, 256]) >>> target = preds * 0.75 >>> metric = StructuralSimilarityIndexMeasure(data_range=1.0) >>> metric.update(preds, target) >>> fig_, ax_ = metric.plot() .. plot:: :scale: 75 >>> # Example plotting multiple values >>> import torch >>> from torchmetrics.image import StructuralSimilarityIndexMeasure >>> preds = torch.rand([3, 3, 256, 256]) >>> target = preds * 0.75 >>> metric = StructuralSimilarityIndexMeasure(data_range=1.0) >>> values = [ ] >>> for _ in range(10): ... values.append(metric(preds, target)) >>> fig_, ax_ = metric.plot(values) _plotrArUrVr:r:rEplot ,r) TrrrNr r!FFNN)__name__ __module__ __qualname____doc__rbool__annotations__rrrfloatrrrrrintrrrJrr<rQrTrrr[ __classcell__r:r:rCrErsf  1          '&rcsFeZdZUdZdZeed<dZeed<dZeed<dZ e ed<d Z e ed <e e ed <e e ed <        d-dedeeeefdee ee fdeddeee ee e ffde de dee dfded d!ed"dffd#d$ Zd e d e d"dfd%d&Zd"e fd'd(Z d.d)eee ee fd*eed"efd+d,ZZS)/*MultiScaleStructuralSimilarityIndexMeasureaF Compute `MultiScaleSSIM`_, Multi-scale Structural Similarity Index Measure. This metric is is a generalization of Structural Similarity Index Measure by incorporating image details at different resolution scores. As input to ``forward`` and ``update`` the metric accepts the following input - ``preds`` (:class:`~torch.Tensor`): Predictions from model - ``target`` (:class:`~torch.Tensor`): Ground truth values As output of `forward` and `compute` the metric returns the following output - ``msssim`` (:class:`~torch.Tensor`): if ``reduction!='none'`` returns float scalar tensor with average MSSSIM value over sample else returns tensor of shape ``(N,)`` with SSIM values per sample Args: gaussian_kernel: If ``True`` (default), a gaussian kernel is used, if false a uniform kernel is used kernel_size: size of the gaussian kernel sigma: Standard deviation of the gaussian kernel reduction: a method to reduce metric score over labels. - ``'elementwise_mean'``: takes the mean - ``'sum'``: takes the sum - ``'none'`` or ``None``: no reduction will be applied data_range: the range of the data. If None, it is determined from the data (max - min). If a tuple is provided then the range is calculated as the difference and input is clamped between the values. The ``data_range`` must be given when ``dim`` is not None. k1: Parameter of structural similarity index measure. k2: Parameter of structural similarity index measure. betas: Exponent parameters for individual similarities and contrastive sensitivities returned by different image resolutions. normalize: When MultiScaleStructuralSimilarityIndexMeasure loss is used for training, it is desirable to use normalizes to improve the training stability. This `normalize` argument is out of scope of the original implementation [1], and it is adapted from https://github.com/jorge-pessoa/pytorch-msssim instead. kwargs: Additional keyword arguments, see :ref:`Metric kwargs` for more info. Return: Tensor with Multi-Scale SSIM score Raises: ValueError: If ``kernel_size`` is not an int or a Sequence of ints with size 2 or 3. ValueError: If ``betas`` is not a tuple of floats with length 2. ValueError: If ``normalize`` is neither `None`, `ReLU` nor `simple`. Example: >>> from torch import rand >>> from torchmetrics.image import MultiScaleStructuralSimilarityIndexMeasure >>> preds = torch.rand([3, 3, 256, 256]) >>> target = preds * 0.75 >>> ms_ssim = MultiScaleStructuralSimilarityIndexMeasure(data_range=1.0) >>> ms_ssim(preds, target) tensor(0.9628) TrrFrrrrrrrrrrNr r!gǺ?g48EG?ga4?g??g9EGr?relur"r$r#r%r&r)r*r+betas. normalize)risimpleNr.r/c  sNtjdi| d} || vrtd| d||dvr)|jdtdddn|jdgdd|jd tdddt|ttfsJtd |t|tret |d vs^t d d |Dsetd|||_ ||_ ||_ ||_||_||_||_t|tstdt|trt dd |Dstd||_| r| dvrtd| |_dS)Nr&r0r1r2r3rr'r4r7zRArgument `kernel_size` expected to be an sequence or an int, or a single int. Got )cs|]}t|tVqdSN)rIre).0ksr:r:rE WzFMultiScaleStructuralSimilarityIndexMeasure.__init__..ztArgument `kernel_size` expected to be an sequence of size 2 or 3 where each element is an int, or a single int. Got z3Argument `betas` is expected to be of a type tuple.csrorp)rIrd)rqbetar:r:rErsgrtz5Argument `betas` is expected to be a tuple of floats.)rirlzNArgument `normalize` to be expected either `None` or one of 'relu' or 'simple'r:)r;r<r=r>r?r@rIrrelenallr"r#r$r%r)r*r+rJrjrk) rAr"r$r#r%r)r*r+rjrkr.rBrCr:rEr<8sD     z3MultiScaleStructuralSimilarityIndexMeasure.__init__c Cst||\}}t|||j|j|j|j|j|j|j|j }|j dvr2t |j t s+td|j |nt |j ts>> # Example plotting a single value >>> from torch import rand >>> from torchmetrics.image import MultiScaleStructuralSimilarityIndexMeasure >>> preds = rand([3, 3, 256, 256]) >>> target = preds * 0.75 >>> metric = MultiScaleStructuralSimilarityIndexMeasure(data_range=1.0) >>> metric.update(preds, target) >>> fig_, ax_ = metric.plot() .. plot:: :scale: 75 >>> # Example plotting multiple values >>> from torch import rand >>> from torchmetrics.image import MultiScaleStructuralSimilarityIndexMeasure >>> preds = rand([3, 3, 256, 256]) >>> target = preds * 0.75 >>> metric = MultiScaleStructuralSimilarityIndexMeasure(data_range=1.0) >>> values = [ ] >>> for _ in range(10): ... values.append(metric(preds, target)) >>> fig_, ax_ = metric.plot(values) rXrZr:r:rEr[r\r) TrrrNr r!rhrir])r^r_r`rarrbrcrrrrdrrrrrerrrrJrr<rQrTrrr[rfr:r:rCrErgsf  <          6rg)collections.abcrtypingrrrrr?rtyping_extensionsr"torchmetrics.functional.image.ssimr r r torchmetrics.metricr torchmetrics.utilities.datar torchmetrics.utilities.importsrtorchmetrics.utilities.plotrr__doctest_skip__rrgr:r:r:rEs      T