o oi@s|ddlmZmZmZmZddlmZddlm Z ddl m Z ddl m Z Gddde ZGdd d e ZGd d d e Zd S) )AnyDictOptionalTuple)random_generator)IntensityAugmentationBase2D)diamond_square)Tensorc seZdZdZ     ddeeefdeeefded ed ed d f fd d Z ddede e efde e e fde ed ef ddZ ZS)RandomPlasmaBrightnessaAdds brightness to the image based on a fractal map generated by the diamond square algorithm. .. image:: _static/img/RandomPlasmaBrightness.png This is based on the original paper: TorMentor: Deterministic dynamic-path, data augmentations with fractals. See: :cite:`tormentor` for more details. .. note:: This function internally uses :func:`kornia.contrib.diamond_square`. Args: roughness: value to scale during the recursion in the generation of the fractal map. intensity: value that scales the intensity values of the generated maps. same_on_batch: apply the same transformation across the batch. p: probability of applying the transformation. keepdim: whether to keep the output shape the same as input (True) or broadcast it to the batch form (False). Examples: >>> rng = torch.manual_seed(0) >>> img = torch.ones(1, 1, 3, 4) >>> RandomPlasmaBrightness(roughness=(0.1, 0.7), p=1.)(img) tensor([[[[0.6415, 1.0000, 0.3142, 0.6836], [1.0000, 0.5593, 0.5556, 0.4566], [0.5809, 1.0000, 0.7005, 1.0000]]]]) g?gffffff??F? roughness intensity same_on_batchpkeepdimreturnNcs6tj||d|dt|dddf|dddf|_dS)Nrrrp_batchrrrsuper__init__rgPlainUniformGenerator_param_generator)selfrrrrr __class__\/home/ubuntu/.local/lib/python3.10/site-packages/kornia/augmentation/_2d/intensity/plasma.pyr7s zRandomPlasmaBrightness.__init__imageparamsflags transformc Csr|j\}}}}|d|} |d|dddd} dt||||f| |j|jdd} | | } || ddS)Nrrdevicedtypershapetoviewrr+r,clamp_) rr#r$r%r&BCHWrrbrightness_mapr!r!r"apply_transformDs $z&RandomPlasmaBrightness.apply_transform)r r FrFN__name__ __module__ __qualname____doc__rfloatboolrr rstrrrr7 __classcell__r!r!rr"r s@    r c s~eZdZdZ    ddeeefdededed d f fd d Z dd ede e efde e e fde ed ef ddZ ZS)RandomPlasmaContrasta@Adds contrast to the image based on a fractal map generated by the diamond square algorithm. .. image:: _static/img/RandomPlasmaContrast.png This is based on the original paper: TorMentor: Deterministic dynamic-path, data augmentations with fractals. See: :cite:`tormentor` for more details. .. note:: This function internally uses :func:`kornia.contrib.diamond_square`. Args: roughness: value to scale during the recursion in the generation of the fractal map. same_on_batch: apply the same transformation across the batch. p: probability of applying the transformation. keepdim: whether to keep the output shape the same as input (True) or broadcast it to the batch form (False). Examples: >>> rng = torch.manual_seed(0) >>> img = torch.ones(1, 1, 3, 4) >>> RandomPlasmaContrast(roughness=(0.1, 0.7), p=1.)(img) tensor([[[[0.9651, 1.0000, 1.0000, 1.0000], [1.0000, 0.9103, 0.8038, 0.9263], [0.6882, 1.0000, 0.9544, 1.0000]]]]) r FrrrrrrNcs,tj||d|dt|dddf|_dS)Nrrrr)rrrrrrr!r"rkszRandomPlasmaContrast.__init__r#r$r%r&c CsT|j\}}}}|d|} dt||||f| |j|jd} |d| dddS)Nrr*rrr()r.r/rr+r,r1) rr#r$r%r&r2r3r4r5r contrast_mapr!r!r"r7us z$RandomPlasmaContrast.apply_transform)r FrFr8r9r!r!rr"rBOs:    rBcseZdZdZ      ddeeefdeeefd eeefd ed ed ed dffdd Z ddede e efde e e fde ed ef ddZ ZS)RandomPlasmaShadowaAdd gaussian noise to a batch of multi-dimensional images. .. image:: _static/img/RandomPlasmaShadow.png This is based on the original paper: TorMentor: Deterministic dynamic-path, data augmentations with fractals. See: :cite:`tormentor` for more details. .. note:: This function internally uses :func:`kornia.contrib.diamond_square`. Args: roughness: value to scale during the recursion in the generation of the fractal map. shade_intensity: value that scales the intensity values of the generated maps. shade_quantity: value to select the pixels to mask. same_on_batch: apply the same transformation across the batch. p: probability of applying the transformation. keepdim: whether to keep the output shape the same as input (True) or broadcast it to the batch form (False). Examples: >>> rng = torch.manual_seed(0) >>> img = torch.ones(1, 1, 3, 4) >>> RandomPlasmaShadow(roughness=(0.1, 0.7), p=1.)(img) tensor([[[[0.7682, 1.0000, 1.0000, 1.0000], [1.0000, 1.0000, 1.0000, 1.0000], [1.0000, 1.0000, 1.0000, 1.0000]]]]) r gr r Frrshade_intensityshade_quantityrrrrNcs@tj||d|dt|dddf|dddf|dddf|_dS)NrrrrGrHr)rrrGrHrrrrr!r"rs      zRandomPlasmaShadow.__init__r#r$r%r&c Cs|j\}}}}|d|} |d|dddd} |d|dddd} t|d||f| |j|jd} | | k|j| } || ddS)NrrGr'r(rHr*rr-) rr#r$r%r&r2_r4r5rrGrH shade_mapr!r!r"r7sz"RandomPlasmaShadow.apply_transform)r rFr FrFr8r9r!r!rr"rE~sF     rEN)typingrrrrkornia.augmentationrr&kornia.augmentation._2d.intensity.baserkornia.contribr kornia.corer r rBrEr!r!r!r"s    5/