o .wi @sddlmZddlmZddlZddlmZmZddlmZddl m Z ddl m Z ddl mZd ed ed eeeffd d Z   dd ed edeedeedeedd ef ddZ   dd ed edeedeedeedd ef ddZdS))Sequence)OptionalN)Tensornn)Literal)_gaussian_kernel_2d)_check_same_shape)reducepredstargetreturncCs`|j|jkrtd|jd|jdt||t|jdkr,td|jd|jd||fS)zUpdate and returns variables required to compute Universal Image Quality Index. Args: preds: Predicted tensor target: Ground truth tensor zEExpected `preds` and `target` to have the same data type. Got preds: z and target: .z@Expected `preds` and `target` to have BxCxHxW shape. Got preds: )dtype TypeErrorrlenshape ValueError)r r r^/home/ubuntu/sommelier/.venv/lib/python3.10/site-packages/torchmetrics/functional/image/uqi.py _uqi_updates  r r?relementwise_mean kernel_sizesigma reduction)rsumnonecCst|dks t|dkrtdt|dt|dtdd|Dr,td|dtdd|Dr=td |d|j}|d }|j}t|||||}|d d d} |d d d} tjj || | | | fd d }tjj || | | | fd d }t ||||||||f} tjj | ||d} | |jd } | d d}| d d}| d | d }t j| d|dd}t j| d|dd}| d|}d|}||}t |jj}d||||||}|d| | | | f}t||S)aCompute Universal Image Quality Index. Args: preds: estimated image target: ground truth image 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 (default) - ``'sum'``: takes the sum - ``'none'`` or ``None``: no reduction will be applied Example: >>> preds = torch.rand([16, 1, 16, 16]) >>> target = preds * 0.75 >>> preds, target = _uqi_update(preds, target) >>> _uqi_compute(preds, target) tensor(0.9216) zOExpected `kernel_size` and `sigma` to have the length of two. Got kernel_size: z and sigma: r css$|] }|ddkp |dkVqdS)r!rNr).0xrrr Qs"z_uqi_compute..z8Expected `kernel_size` to have odd positive number. Got css|]}|dkVqdS)rNr)r"yrrrr$Tsz.Expected `sigma` to have positive number. Got rreflect)mode)groupsg)minr.)rranydevicesizerrr functionalpadtorchcatconv2dsplitrpowclampfinfoepsr )r r rrrr-channelrkernelpad_hpad_w input_listoutputs output_list mu_pred_sq mu_target_sqmu_pred_target sigma_pred_sqsigma_target_sqsigma_pred_targetupperlowerr8uqi_idxrrr _uqi_compute/sF    rIcCst||\}}t|||||S)aUniversal Image Quality Index. Args: preds: estimated image target: ground truth image 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 (default) - ``'sum'``: takes the sum - ``'none'`` or ``None``: no reduction will be applied Return: Tensor with UniversalImageQualityIndex score Raises: TypeError: If ``preds`` and ``target`` don't have the same data type. ValueError: If ``preds`` and ``target`` don't have ``BxCxHxW shape``. ValueError: If the length of ``kernel_size`` or ``sigma`` is not ``2``. ValueError: If one of the elements of ``kernel_size`` is not an ``odd positive number``. ValueError: If one of the elements of ``sigma`` is not a ``positive number``. Example: >>> from torchmetrics.functional.image import universal_image_quality_index >>> preds = torch.rand([16, 1, 16, 16]) >>> target = preds * 0.75 >>> universal_image_quality_index(preds, target) tensor(0.9216) References: [1] Zhou Wang and A. C. Bovik, "A universal image quality index," in IEEE Signal Processing Letters, vol. 9, no. 3, pp. 81-84, March 2002, doi: 10.1109/97.995823. [2] Zhou Wang, A. C. Bovik, H. R. Sheikh and E. P. Simoncelli, "Image quality assessment: from error visibility to structural similarity," in IEEE Transactions on Image Processing, vol. 13, no. 4, pp. 600-612, April 2004, doi: 10.1109/TIP.2003.819861. )rrI)r r rrrrrruniversal_image_quality_indexws3rJ)rrr)collections.abcrtypingrr1rrtyping_extensionsr#torchmetrics.functional.image.utilsrtorchmetrics.utilities.checksr"torchmetrics.utilities.distributedr tuplerintfloatrIrJrrrrsR       K