i ddlZddlmZddlmZddlmZddlmZdededejd ejd ef d Z d ed eded efdZ dd ed edededd ef dZ y)N)Tensor)conv2d)Literal) dim_zero_catwin_sizesigmadtypedevicereturnctj||||dz dz z }|dz}tj|jd|jdz d|dzzz }|tj|z}|S)Nr r r@)torcharangeexp unsqueezesum)rrr r coordsgs v/Volumes/fast/ai/experiments/voice-extract-mac/.venv/lib/python3.12/site-packages/torchmetrics/functional/image/vif.py_filterrsx \\(% ?8atT1}M 19FF+Aq#A#ssN;F5&)!Q!SqS.9E66* %( !| k #h.++fVQY&?,&NTWXVE1Hf%= %KQTU#FUNF;oM /C"7 8#a*<&<< $$)$/ 4 !#$ 41u)$/ 4$[[5 UYYu{{3!S&O1Kz\fOf1g+ghnwxx eii C/J2N,N OU^__ IL z !!r reductionmeannonec d|jddks|jddkr0td|jdd|jdd|jddks|jddkr0td|jdd|jdd|j|jk7r&td|jd |jd |d vrtd |t|jd Dcgc])}t |dd|ddddf|dd|ddddf|+}}t |jd d kDr&t j|djdn|d}|dk(r|jS|Scc}w)aCompute Pixel-Based Visual Information Fidelity (VIF-P). VIF is a full-reference metric that measures the amount of visual information preserved in a distorted image compared to the reference image. Args: preds: Predicted images of shape (N, C, H, W). Height and width must be at least 41. target: Ground truth images of shape (N, C, H, W). Must match preds in shape. sigma_n_sq: Variance of the visual noise. Default: 2.0. reduction: Method for reducing the metric across the batch. - "mean": Return a tensor average over the batch. - "none": Return a VIF score for each sample as a 1D tensor of shape (N,). Returns: torch.Tensor: VIF score(s). The shape depends on the `reduction` argument: - If ``reduction="mean"``, returns a scalar tensor. - If ``reduction="none"``, returns a tensor of shape ``(N,)``. Raises: ValueError: If input dimensions are smaller than ``41x41``. ValueError: If ``preds`` and ``target`` shapes don't match. ValueError: If ``reduction`` is not ``"mean"`` or ``"none"``. Example: >>> from torchmetrics.functional.image import visual_information_fidelity >>> preds = torch.randn(4, 3, 41, 41, generator=torch.Generator().manual_seed(42)) >>> target = torch.randn(4, 3, 41, 41, generator=torch.Generator().manual_seed(43)) >>> visual_information_fidelity(preds, target, reduction="none") tensor([0.0040, 0.0049, 0.0017, 0.0039]) )z8Invalid size of preds. Expected at least 41x41, but got x!z9Invalid size of target. Expected at least 41x41, but got z7`preds` and `target` must have the same shape, but got z vs .r=z7Argument `reduction` must be 'mean' or 'none', but got rNrr#r>) r' ValueErrorshaper(r;rrstackr>)rrrr<iper_channel_scoresvif_per_samples rvisual_information_fidelityrMYsR zz"~ejjnr1STYT^T^_aTbSccdejeoeopresdttuvww {{2v{{23G TVGXXYZ`ZeZefhZiYjjk l   {{fll"RSXS^S^R__cdjdpdpcqqrstt((RS\R]^__V[[`[e[efg[hUiUiPQq!Qz*F1aA:,> KUi":?**Q-!:K &A.33A6QcdeQfNF""$$ s.F-)rr>)rrtorch.nn.functionalrtyping_extensionsrtorchmetrics.utilities.datarfloatr r rr;rMrrrSs &%4  e  E  %++  u||  X^  4"F4"F4"4"&4"t)/ A A AA~& A  Ar