o oi@@sNddlmZddlmZddlmZmZmZddlZddl m Z m Z m Z ddl mZddl mZmZmZddlmZmZmZdd lmZdd lmZmZdlddZdlddZdlddZdlddZdlddZ dmdnddZ!dodpd#d$Z"dld%d&Z#dodpd'd(Z$dodpd)d*Z%dqdrd1d2Z&dsdtd4d5Z'dudvd7d8Z( + dwdxd;d<Z)edyd?d@Z*edzdAdBZ+d{dEdFZ,d|dIdJZ-d}dLdMZ.ed~dNdOZ/ed~dPdQZ0dddTdUZ1GdVdWdWeZ2GdXdYdYeZ3GdZd[d[eZ4Gd\d]d]eZ5Gd^d_d_eZ6Gd`dadaeZ7GdbdcdceZ8GdddedeeZ9GdfdgdgeZ:GdhdidieZ;GdjdkdkeZ>> x = torch.ones(1, 3, 3, 3) >>> adjust_saturation_with_gray_subtraction(x, 2.).shape torch.Size([1, 3, 3, 3]) >>> x = torch.ones(2, 3, 3, 3) >>> y = torch.tensor([1., 2.]) >>> adjust_saturation_with_gray_subtraction(x, y).shape torch.Size([2, 3, 3, 3]) rrz$Image should be an RGB or gray imagerr#rr?)rrr(r)r rr.r*r+rrr,r-rr0)rrx_other x_adjustedr6r7r7r8'adjust_saturation_with_gray_subtractionCs    r>cCt|}t||}t|}|S)aAdjust color saturation of an image. .. image:: _static/img/adjust_saturation.png The image is expected to be an RGB image in the range of [0, 1]. Args: image: Image/Tensor to be adjusted in the shape of :math:`(*, 3, H, W)`. factor: How much to adjust the saturation. 0 will give a black and white image, 1 will give the original image while 2 will enhance the saturation by a factor of 2. saturation_mode: The mode to adjust saturation. Return: Adjusted image in the shape of :math:`(*, 3, H, W)`. .. note:: See a working example `here `__. Example: >>> x = torch.ones(1, 3, 3, 3) >>> adjust_saturation(x, 2.).shape torch.Size([1, 3, 3, 3]) >>> x = torch.ones(2, 3, 3, 3) >>> y = torch.tensor([1., 2.]) >>> adjust_saturation(x, y).shape torch.Size([2, 3, 3, 3]) )r r9rrrx_hsvr=r6r7r7r8adjust_saturation|s rBcCst|dtt|ttfdt|t|trt|}||j |j }t |j t |j kr>|d}t |j t |j ks0tj |ddd\}}}dt}t|||}tj|||gdd}|S) zNAdjust hue of an image. Expecting image to be in hsv format already. rzRThe factor should be a float number or Tensor in the range between [-PI, PI]. Got rrrr r')rrr(r)r typer*r+r,rrr-r.r/rfmodr1)rrr2r3r4divisorh_outr6r7r7r8adjust_hue_raws     rHcCr?)a Adjust hue of an image. .. image:: _static/img/adjust_hue.png The image is expected to be an RGB image in the range of [0, 1]. Args: image: Image to be adjusted in the shape of :math:`(*, 3, H, W)`. factor: How much to shift the hue channel. Should be in [-PI, PI]. PI and -PI give complete reversal of hue channel in HSV space in positive and negative direction respectively. 0 means no shift. Therefore, both -PI and PI will give an image with complementary colors while 0 gives the original image. Return: Adjusted image in the shape of :math:`(*, 3, H, W)`. .. note:: See a working example `here `__. Example: >>> x = torch.ones(1, 3, 2, 2) >>> adjust_hue(x, 3.141516).shape torch.Size([1, 3, 2, 2]) >>> x = torch.ones(2, 3, 3, 3) >>> y = torch.ones(2) * 3.141516 >>> adjust_hue(x, y).shape torch.Size([2, 3, 3, 3]) )r rHrr@r7r7r8 adjust_hues rIr;inputgammagaincCsZt|tstdt|t|ttfstdt|t|ttfs.tdt|t|tr8t|g}t|trBt|g}||j|j}||j|j}|dkrct d||dkrpt d|t t |j t |j D] }t j|dd}q|t t |j t |j D] }t j|dd}q|t ||}t |dd }|S) aPerform gamma correction on an image. .. image:: _static/img/adjust_contrast.png The input image is expected to be in the range of [0, 1]. Args: input: Image to be adjusted in the shape of :math:`(*, H, W)`. gamma: Non negative real number, same as y\gammay in the equation. gamma larger than 1 make the shadows darker, while gamma smaller than 1 make dark regions lighter. gain: The constant multiplier. Return: Adjusted image in the shape of :math:`(*, H, W)`. .. note:: See a working example `here `__. Example: >>> x = torch.ones(1, 1, 2, 2) >>> adjust_gamma(x, 1.0, 2.0) tensor([[[[1., 1.], [1., 1.]]]]) >>> x = torch.ones(2, 5, 3, 3) >>> y1 = torch.ones(2) * 1.0 >>> y2 = torch.ones(2) * 2.0 >>> adjust_gamma(x, y1, y2).shape torch.Size([2, 5, 3, 3]) Input type is not a Tensor. Got z4The gamma should be a positive float or Tensor. Got z3The gain should be a positive float or Tensor. Got r:z Gamma must be non-negative. Got zGain must be non-negative. Got r'r;)r(r TypeErrorrDr)r,rrany ValueErrorranger-r.r* unsqueezepowr0)rJrKrL_x_adjustr6r7r7r8 adjust_gammas. !      rWT clip_outputboolcCst|dtt|ttfdt|trtj||j|jd}n t|tr,| |j|j}t |j t |j krD|d}t |j t |j ks6tt |dkd||}|rZ|j ddd }|S) aAdjust the contrast of an image tensor. .. image:: _static/img/adjust_contrast.png This implementation follows Szeliski's book convention, where contrast is defined as a `multiplicative` operation directly to raw pixel values. Beware that other frameworks might use different conventions which can be difficult to reproduce exact results. The input image and factor is expected to be in the range of [0, 1]. .. tip:: This is not the preferred way to adjust the contrast of an image. Ideally one must implement :func:`kornia.enhance.adjust_gamma`. More details in the following link: https://scikit-image.org/docs/dev/auto_examples/color_exposure/plot_log_gamma.html#sphx-glr-auto-examples-color-exposure-plot-log-gamma-py Args: image: Image to be adjusted in the shape of :math:`(*, H, W)`. factor: Contrast adjust factor per element in the batch. 0 generates a completely black image, 1 does not modify the input image while any other non-negative number modify the brightness by this factor. clip_output: whether to clip the output image with range of [0, 1]. Return: Adjusted image in the shape of :math:`(*, H, W)`. .. note:: See a working example `here `__. Example: >>> import torch >>> x = torch.ones(1, 1, 2, 2) >>> adjust_contrast(x, 0.5) tensor([[[[0.5000, 0.5000], [0.5000, 0.5000]]]]) >>> x = torch.ones(2, 5, 3, 3) >>> y = torch.tensor([0.65, 0.50]) >>> adjust_contrast(x, y).shape torch.Size([2, 5, 3, 3]) rrrrrz!Contrast factor must be positive.r:r;r$)rrr(r)r r*r+rrr,r-r.rPr0rrrX img_adjustr7r7r8adjust_contrast:s +  r\cCst|dtt|ttfdt|trtj||j|jd}n t|tr,| |j|j}t |j t |j krD|d}t |j t |j ks6|j ddkrTt | dd}n| }|||d |}|jd d d }|S) a!Adjust the contrast of an image tensor by subtracting the mean over channels. .. note:: this is just a convenience function to have compatibility with Pil. For exact definition of image contrast adjustment consider using :func:`kornia.enhance.adjust_gamma`. Args: image: Image to be adjusted in the shape of :math:`(*, H, W)`. factor: Contrast adjust factor per element in the batch. 0 generates a completely black image, 1 does not modify the input image while any other non-negative number modify the brightness by this factor. Return: Adjusted image in the shape of :math:`(*, H, W)`. Example: >>> import torch >>> x = torch.ones(1, 1, 2, 2) >>> adjust_contrast_with_mean_subtraction(x, 0.5) tensor([[[[1., 1.], [1., 1.]]]]) >>> x = torch.ones(2, 5, 3, 3) >>> y = torch.tensor([0.65, 0.50]) >>> adjust_contrast_with_mean_subtraction(x, y).shape torch.Size([2, 5, 3, 3]) rrrrrr)rNTr#r:r;r$)rrr(r)r r*r+rrr,r-r.rmeanr0)rrimg_meanr[r7r7r8%adjust_contrast_with_mean_subtraction~s   r`cCst|dtt|ttfdt|trtj||j|jd}n t|tr,| |j|j}t |j t |j krD|d}t |j t |j ks6||}|rQ|j ddd}|S)aOAdjust the brightness of an image tensor. .. image:: _static/img/adjust_brightness.png This implementation follows Szeliski's book convention, where brightness is defined as an `additive` operation directly to raw pixel and shift its values according the applied factor and range of the image values. Beware that other framework might use different conventions which can be difficult to reproduce exact results. The input image and factor is expected to be in the range of [0, 1]. .. tip:: By applying a large factor might prouce clipping or loss of image detail. We recommenda to apply small factors to avoid the mentioned issues. Ideally one must implement the adjustment of image intensity with other techniques suchs as :func:`kornia.enhance.adjust_gamma`. More details in the following link: https://scikit-image.org/docs/dev/auto_examples/color_exposure/plot_log_gamma.html#sphx-glr-auto-examples-color-exposure-plot-log-gamma-py Args: image: Image to be adjusted in the shape of :math:`(*, H, W)`. factor: Brightness adjust factor per element in the batch. It's recommended to bound the factor by [0, 1]. 0 does not modify the input image while any other number modify the brightness. clip_output: Whether to clip output to be in [0,1]. Return: Adjusted tensor in the shape of :math:`(*, H, W)`. .. note:: See a working example `here `__. Example: >>> x = torch.ones(1, 1, 2, 2) >>> adjust_brightness(x, 1.) tensor([[[[1., 1.], [1., 1.]]]]) >>> x = torch.ones(2, 5, 3, 3) >>> y = torch.tensor([0.25, 0.50]) >>> adjust_brightness(x, y).shape torch.Size([2, 5, 3, 3]) rrrrr:r;r$ rrr(r)r r*r+rrr,r-r.r0rZr7r7r8adjust_brightnesss ,  rbcCst|dtt|ttfdt|trtj||j|jd}n t|tr,| |j|j}t |j t |j krD|d}t |j t |j ks6||}|rQ|j ddd}|S)aAdjust the brightness accumulatively of an image tensor. This implementation follows PIL convention. The input image and factor is expected to be in the range of [0, 1]. Args: image: Image to be adjusted in the shape of :math:`(*, H, W)`. factor: Brightness adjust factor per element in the batch. It's recommended to bound the factor by [0, 1]. 0 does not modify the input image while any other number modify the brightness. clip_output: Whether to clip output to be in [0,1]. Return: Adjusted tensor in the shape of :math:`(*, H, W)`. Example: >>> x = torch.ones(1, 1, 2, 2) >>> adjust_brightness_accumulative(x, 1.) tensor([[[[1., 1.], [1., 1.]]]]) >>> x = torch.ones(2, 5, 3, 3) >>> y = torch.tensor([0.25, 0.50]) >>> adjust_brightness_accumulative(x, y).shape torch.Size([2, 5, 3, 3]) rrrrr:r;r$rarZr7r7r8adjust_brightness_accumulatives   rc? Fcutoffr)invcCsJt|d|rddd|||}|Sdd|||}|S)a$Adjust sigmoid correction on the input image tensor. The input image is expected to be in the range of [0, 1]. Reference: [1]: Gustav J. Braun, "Image Lightness Rescaling Using Sigmoidal Contrast Enhancement Functions", http://markfairchild.org/PDFs/PAP07.pdf Args: image: Image to be adjusted in the shape of :math:`(*, H, W)`. cutoff: The cutoff of sigmoid function. gain: The multiplier of sigmoid function. inv: If is set to True the function will return the inverse sigmoid correction. Returns: Adjusted tensor in the shape of :math:`(*, H, W)`. Example: >>> x = torch.ones(1, 1, 2, 2) >>> adjust_sigmoid(x, gain=0) tensor([[[[0.5000, 0.5000], [0.5000, 0.5000]]]]) rr#)rexp)rrfrLrgr[r7r7r8adjust_sigmoid1s rir#cCsFt|d|rd|d|}nd||}|r!|jddd}|S)aAdjust log correction on the input image tensor. The input image is expected to be in the range of [0, 1]. Reference: [1]: http://www.ece.ucsb.edu/Faculty/Manjunath/courses/ece178W03/EnhancePart1.pdf Args: image: Image to be adjusted in the shape of :math:`(*, H, W)`. gain: The multiplier of logarithmic function. inv: If is set to True the function will return the inverse logarithmic correction. clip_output: Whether to clip the output image with range of [0, 1]. Returns: Adjusted tensor in the shape of :math:`(*, H, W)`. Example: >>> x = torch.zeros(1, 1, 2, 2) >>> adjust_log(x, inv=True) tensor([[[[0., 0.], [0., 0.]]]]) rrCr#r:r;r$)rlog2r0)rrLrgrXr[r7r7r8 adjust_logSs rk thresholdscsttstdtt|ttfstdt|t|tr]t|jdkr]dt|kr:t|jdksGtddd|| j  j }t fdd|D}t |kd S) aFor each pixel in the image, select the pixel if the value is less than the threshold. Otherwise, subtract 1.0 from the pixel. Args: input: image or batched images to solarize. thresholds: solarize thresholds. If int or one element tensor, input will be solarized across the whole batch. If 1-d tensor, input will be solarized element-wise, len(thresholds) == len(input). Returns: Solarized images. rM3The factor should be either a float or Tensor. Got rr#z*thresholds must be a 1-d vector of shape (,). Got c g|] }|jjddqSrNexpandr..0xrJr7r8  z_solarize..r;)r(r rOrDr)r-r.sizeAssertionErrorr,rrr*stackwhere)rJrlr7rvr8 _solarizeys  r} additionsOptional[Union[float, Tensor]]csNttstdtt|ttfstdt|t|tr(t|}|durt|ttfs>> x = torch.rand(1, 4, 3, 3) >>> out = solarize(x, thresholds=0.5, additions=0.) >>> out.shape torch.Size([1, 4, 3, 3]) >>> x = torch.rand(2, 4, 3, 3) >>> thresholds = torch.tensor([0.8, 0.5]) >>> additions = torch.tensor([-0.25, 0.25]) >>> solarize(x, thresholds, additions).shape torch.Size([2, 4, 3, 3]) rMrmNrdgz5The value of 'addition' is between -0.5 and 0.5. Got .rr#z)additions must be a 1-d vector of shape (rncrorprqrsrvr7r8rwrxzsolarize..r:r;)r(r rOrDr)r*r+allrzr-r.ryr,rrr{r0r})rJrlr~r7rvr8solarizes* &       rbitsUnion[int, Tensor]cst|tstdt|t|ttfstdt|t|tr(t|}ddddd d dfd d }t|jdksNt|jdkrSt|dkrS|||Sg}t|jdkr|jd|jdkrwt d|jdd|jddt |jdD]}| |||||q~tj |ddS|j|jdt|jkrt d|jd|jdt|jd|j dg|jt|jdR}|}t |jdD]}| |||||qtj |ddj|jS)aReduce the number of bits for each color channel. .. image:: _static/img/posterize.png Non-differentiable function, ``torch.uint8`` involved. Args: input: image tensor with shape :math:`(*, C, H, W)` to posterize. bits: number of high bits. Must be in range [0, 8]. If int or one element tensor, input will be posterized by this bits. If 1-d tensor, input will be posterized element-wisely, len(bits) == input.shape[-3]. If n-d tensor, input will be posterized element-channel-wisely, bits.shape == input.shape[:len(bits.shape)] Returns: Image with reduced color channels with shape :math:`(*, C, H, W)`. Example: >>> x = torch.rand(1, 6, 3, 3) >>> out = posterize(x, bits=8) >>> torch.testing.assert_close(x, out) >>> x = torch.rand(2, 6, 3, 3) >>> bits = torch.tensor([4, 2]) >>> posterize(x, bits).shape torch.Size([2, 6, 3, 3]) rMz'bits type is not an int or Tensor. Got rJr shiftrcSs$|dtjd||jdSNrCo@r,r*uint8rrJrr7r7r8 _left_shift $zposterize.._left_shiftcSs$|dtjd||jdSrrrr7r7r8 _right_shiftrzposterize.._right_shiftrcs:|dkr t|S|dkr|Sd|}|||S)Nr)r* zeros_likeclone)rJrrrr7r8_posterize_ones  z!posterize.._posterize_onerr#z5Batch size must be equal between bits and input. Got z, rr'NzApply sharpness to the input tensor. .. image:: _static/img/sharpness.png Implemented Sharpness function from PIL using torch ops. This implementation refers to: https://github.com/tensorflow/tpu/blob/master/models/official/efficientnet/autoaugment.py#L326 Args: input: image tensor with shape :math:`(*, C, H, W)` to sharpen. factor: factor of sharpness strength. Must be above 0. If float or one element tensor, input will be sharpened by the same factor across the whole batch. If 1-d tensor, input will be sharpened element-wisely, len(factor) == len(input). Returns: Sharpened image or images with shape :math:`(*, C, H, W)`. Example: >>> x = torch.rand(1, 1, 5, 5) >>> sharpness(x, 0.5).shape torch.Size([1, 1, 5, 5]) rrzSInput batch size shall match with factor size if factor is not a 0-dim tensor. Got z and )r#r#r#)r#r#)rrr#r N)biasstridegroupsr:r;)r#r#r#r#cs$g|]}t|||qSr7) _blend_onertrrrJresultr7r8rwks$zsharpness..)r(r r*r+rrr-ryr.Sizerzrrepeatnn functionalconv2dr0 ones_likepadr|rr{rR)rJrkernel degeneratemask padded_maskpadded_degenerater7rr8 sharpness6s2 ($    $rinput1input2cCst|ts td|dt|tstd|dt|tr/t|dkr/td|d|dkr5|S|dkr;|S|||}||}|dkrO|dkrO|St|ddS)a@Blend two images into one. Args: input1: image tensor with shapes like :math:`(H, W)` or :math:`(D, H, W)`. input2: image tensor with shapes like :math:`(H, W)` or :math:`(D, H, W)`. factor: factor 0-dim tensor. Returns: : image tensor with the batch in the zero position. z`input1` must be a tensor. Got rrz6Factor shall be a float or single element tensor. Got r:r;r#)r(r rzr-ryr*r0)rrrdiffrr7r7r8rns   rhistostepcCs`tj|ddd}tjt|d||dd}ttjd|j|jd|ddg}t|ddS) NrCtrunc rounding_moderr#rrNr)r*divcumsumr1zerosrrr0)rr step_trunclutr7r7r8 _build_luts&rimcCs0|}|}|dkr$t|tjd|jds$td|d|dkr@t|tjd|jds@td|dt|j }|dvrQt d|d|d }t |d d d d }t ||d kdg}tj t||dd dd}|d kr|}|d Stt||d |}||}|d S)zScale the data in the channel to implement equalize. Args: im: image tensor with shapes like :math:`(H, W)` or :math:`(D, H, W)`. Returns: image tensor with the batch in the zero position. r:)rz?Values in the input tensor must greater or equal to 0.0. Found rr;z=Values in the input tensor must lower or equal to 1.0. Found )rCrz0Input tensor must have 2 or 3 dimensions. Found rrr)binsr%r&rNrr)r%r&itemr*iscloser+rrQr-r.rOrrrsumgatherrrlong reshape_as)rmin_max_ndimsr nonzero_historrr7r7r8_scale_channels& $$  rcBg}|D]tfddttD}||qt|S)aAApply equalize on the input tensor. .. image:: _static/img/equalize.png Implements Equalize function from PIL using PyTorch ops based on uint8 format: https://github.com/tensorflow/tpu/blob/5f71c12a020403f863434e96982a840578fdd127/models/official/efficientnet/autoaugment.py#L355 Args: input: image tensor to equalize with shape :math:`(*, C, H, W)`. Returns: Equalized image tensor with shape :math:`(*, C, H, W)`. Example: >>> x = torch.rand(1, 2, 3, 3) >>> equalize(x).shape torch.Size([1, 2, 3, 3]) cs&g|]}t|ddddfqSNrrrr7r8rws&zequalize..r*r{rRr-r)rJr scaled_imager7rr8equalizes    rcr)aEqualize the values for a 3D volumetric tensor. Implements Equalize function for a sequence of images using PyTorch ops based on uint8 format: https://github.com/tensorflow/tpu/blob/master/models/official/efficientnet/autoaugment.py#L352 Args: input: image tensor with shape :math:`(*, C, D, H, W)` to equalize. Returns: Equalized volume with shape :math:`(B, C, D, H, W)`. c s,g|]}t|ddddddfqSrrrvolumer7r8rws,zequalize3d..r)rJr scaled_inputr7rr8 equalize3ds    rmax_valOptional[Tensor]cCs^t|tstdtt|durtdg}n|}t|ts(tdt||||S)aInvert the values of an input image tensor by its maximum value. .. image:: _static/img/invert.png Args: image: The input tensor to invert with an arbitatry shape. max_val: The expected maximum value in the input tensor. The shape has to according to the input tensor shape, or at least has to work with broadcasting. Example: >>> img = torch.rand(1, 2, 4, 4) >>> invert(img).shape torch.Size([1, 2, 4, 4]) >>> img = 255. * torch.rand(1, 2, 3, 4, 4) >>> invert(img, torch.as_tensor(255.)).shape torch.Size([1, 2, 3, 4, 4]) >>> img = torch.rand(1, 3, 4, 4) >>> invert(img, torch.as_tensor([[[[1.]]]])).shape torch.Size([1, 3, 4, 4]) zInput is not a Tensor. Got: Nr;zmax_val is not a Tensor. Got: )r(r rzrDrJr r,)rr_max_valr7r7r8inverts   rcNeZdZUdZgdZded<gdZded<dfd d ZdddZZ S)AdjustSaturationazAdjust color saturation of an image. The input image is expected to be an RGB image in the range of [0, 1]. Args: saturation_factor: How much to adjust the saturation. 0 will give a black and white image, 1 will give the original image while 2 will enhance the saturation by a factor of 2. saturation_mode: The mode to adjust saturation. Shape: - Input: Image/Tensor to be adjusted in the shape of :math:`(*, 3, H, W)`. - Output: Adjusted image in the shape of :math:`(*, 3, H, W)`. Example: >>> x = torch.ones(1, 3, 3, 3) >>> AdjustSaturation(2.)(x) tensor([[[[1., 1., 1.], [1., 1., 1.], [1., 1., 1.]], [[1., 1., 1.], [1., 1., 1.], [1., 1., 1.]], [[1., 1., 1.], [1., 1., 1.], [1., 1., 1.]]]]) >>> x = torch.ones(2, 3, 3, 3) >>> y = torch.ones(2) >>> out = AdjustSaturation(y)(x) >>> torch.nn.functional.mse_loss(x, out) tensor(0.) rNrrNrNClassVar[list[int]]ONNX_DEFAULT_INPUTSHAPEONNX_DEFAULT_OUTPUTSHAPEsaturation_factorrrNonect||_dSrsuper__init__rselfr __class__r7r8rF  zAdjustSaturation.__init__rJr cC t||jSr)rBrrrJr7r7r8forwardJ zAdjustSaturation.forwardrrrrrJr rr __name__ __module__ __qualname____doc__r__annotations__rrr __classcell__r7r7rr8rs $rcr)#AdjustSaturationWithGraySubtractiona6Adjust color saturation of an image. This implementation aligns PIL. Hence, the output is close to TorchVision. The input image is expected to be in the range of [0, 1]. The input image is expected to be an RGB or gray image in the range of [0, 1]. Args: saturation_factor: How much to adjust the saturation. 0 will give a black and white image, 1 will give the original image while 2 will enhance the saturation by a factor of 2. saturation_mode: The mode to adjust saturation. Shape: - Input: Image/Tensor to be adjusted in the shape of :math:`(*, 3, H, W)`. - Output: Adjusted image in the shape of :math:`(*, 3, H, W)`. Example: >>> x = torch.ones(1, 3, 3, 3) >>> AdjustSaturationWithGraySubtraction(2.)(x) tensor([[[[1., 1., 1.], [1., 1., 1.], [1., 1., 1.]], [[1., 1., 1.], [1., 1., 1.], [1., 1., 1.]], [[1., 1., 1.], [1., 1., 1.], [1., 1., 1.]]]]) >>> x = torch.ones(2, 3, 3, 3) >>> y = torch.ones(2) >>> out = AdjustSaturationWithGraySubtraction(y)(x) >>> torch.nn.functional.mse_loss(x, out) tensor(0.) rrrrrrrrcrrrrrr7r8ryrz,AdjustSaturationWithGraySubtraction.__init__rJr cCrr)r>rrr7r7r8r}rz+AdjustSaturationWithGraySubtraction.forwardrrrr7r7rr8rN 'rcr) AdjustHuea9Adjust hue of an image. This implementation aligns PIL. Hence, the output is close to TorchVision. The input image is expected to be in the range of [0, 1]. The input image is expected to be an RGB image in the range of [0, 1]. Args: hue_factor: How much to shift the hue channel. Should be in [-PI, PI]. PI and -PI give complete reversal of hue channel in HSV space in positive and negative direction respectively. 0 means no shift. Therefore, both -PI and PI will give an image with complementary colors while 0 gives the original image. Shape: - Input: Image/Tensor to be adjusted in the shape of :math:`(*, 3, H, W)`. - Output: Adjusted image in the shape of :math:`(*, 3, H, W)`. Example: >>> x = torch.ones(1, 3, 3, 3) >>> AdjustHue(3.141516)(x) tensor([[[[1., 1., 1.], [1., 1., 1.], [1., 1., 1.]], [[1., 1., 1.], [1., 1., 1.], [1., 1., 1.]], [[1., 1., 1.], [1., 1., 1.], [1., 1., 1.]]]]) >>> x = torch.ones(2, 3, 3, 3) >>> y = torch.ones(2) * 3.141516 >>> AdjustHue(y)(x).shape torch.Size([2, 3, 3, 3]) rrrr hue_factorrrrcrr)rrr)rrrr7r8rrzAdjustHue.__init__rJr cCrr)rIrrr7r7r8rrzAdjustHue.forward)rrrrrrr7r7rr8rrrcs.eZdZdZddfdd Zdd d ZZS) AdjustGammaajPerform gamma correction on an image. The input image is expected to be in the range of [0, 1]. Args: gamma: Non negative real number, same as y\gammay in the equation. gamma larger than 1 make the shadows darker, while gamma smaller than 1 make dark regions lighter. gain: The constant multiplier. Shape: - Input: Image to be adjusted in the shape of :math:`(*, N)`. - Output: Adjusted image in the shape of :math:`(*, N)`. Example: >>> x = torch.ones(1, 1, 3, 3) >>> AdjustGamma(1.0, 2.0)(x) tensor([[[[1., 1., 1.], [1., 1., 1.], [1., 1., 1.]]]]) >>> x = torch.ones(2, 5, 3, 3) >>> y1 = torch.ones(2) * 1.0 >>> y2 = torch.ones(2) * 2.0 >>> AdjustGamma(y1, y2)(x).shape torch.Size([2, 5, 3, 3]) r;rKrrLrrcst||_||_dSr)rrrKrL)rrKrLrr7r8rs  zAdjustGamma.__init__rJr cCst||j|jSr)rWrKrLrr7r7r8rszAdjustGamma.forwardr;)rKrrLrrrrrrrrrrrr7r7rr8rsrc,eZdZdZd fdd Zd d d ZZS)AdjustContrastaAdjust Contrast of an image. This implementation aligns OpenCV, not PIL. Hence, the output differs from TorchVision. The input image is expected to be in the range of [0, 1]. Args: contrast_factor: Contrast adjust factor per element in the batch. 0 generates a completely black image, 1 does not modify the input image while any other non-negative number modify the brightness by this factor. Shape: - Input: Image/Input to be adjusted in the shape of :math:`(*, N)`. - Output: Adjusted image in the shape of :math:`(*, N)`. Example: >>> x = torch.ones(1, 1, 3, 3) >>> AdjustContrast(0.5)(x) tensor([[[[0.5000, 0.5000, 0.5000], [0.5000, 0.5000, 0.5000], [0.5000, 0.5000, 0.5000]]]]) >>> x = torch.ones(2, 5, 3, 3) >>> y = torch.ones(2) >>> AdjustContrast(y)(x).shape torch.Size([2, 5, 3, 3]) contrast_factorrrrcrrrrrrrrr7r8rrzAdjustContrast.__init__rJr cCrr)r\rrr7r7r8rrzAdjustContrast.forwardrrrrrrr7r7rr8rsrcr)!AdjustContrastWithMeanSubtractionaAdjust Contrast of an image. This implementation aligns PIL. Hence, the output is close to TorchVision. The input image is expected to be in the range of [0, 1]. Args: contrast_factor: Contrast adjust factor per element in the batch by subtracting its mean grayscaled version. 0 generates a completely black image, 1 does not modify the input image while any other non-negative number modify the brightness by this factor. Shape: - Input: Image/Input to be adjusted in the shape of :math:`(*, N)`. - Output: Adjusted image in the shape of :math:`(*, N)`. Example: >>> x = torch.ones(1, 1, 3, 3) >>> AdjustContrastWithMeanSubtraction(0.5)(x) tensor([[[[1., 1., 1.], [1., 1., 1.], [1., 1., 1.]]]]) >>> x = torch.ones(2, 5, 3, 3) >>> y = torch.ones(2) >>> AdjustContrastWithMeanSubtraction(y)(x).shape torch.Size([2, 5, 3, 3]) rrrrcrrrrrr7r8r rz*AdjustContrastWithMeanSubtraction.__init__rJr cCrr)r`rrr7r7r8r$rz)AdjustContrastWithMeanSubtraction.forwardr rrr7r7rr8r sr cr)AdjustBrightnessaaAdjust Brightness of an image. This implementation aligns OpenCV, not PIL. Hence, the output differs from TorchVision. The input image is expected to be in the range of [0, 1]. Args: brightness_factor: Brightness adjust factor per element in the batch. 0 does not modify the input image while any other number modify the brightness. Shape: - Input: Image/Input to be adjusted in the shape of :math:`(*, N)`. - Output: Adjusted image in the shape of :math:`(*, N)`. Example: >>> x = torch.ones(1, 1, 3, 3) >>> AdjustBrightness(1.)(x) tensor([[[[1., 1., 1.], [1., 1., 1.], [1., 1., 1.]]]]) >>> x = torch.ones(2, 5, 3, 3) >>> y = torch.ones(2) >>> AdjustBrightness(y)(x).shape torch.Size([2, 5, 3, 3]) brightness_factorrrrcrrrrr rr rr7r8rErzAdjustBrightness.__init__rJr cCrr)rbr rr7r7r8rIrzAdjustBrightness.forwardr rrrrrr7r7rr8r (r c.eZdZdZddfd d ZdddZZS) AdjustSigmoidaAdjust the contrast of an image tensor or performs sigmoid correction on the input image tensor. The input image is expected to be in the range of [0, 1]. Reference: [1]: Gustav J. Braun, "Image Lightness Rescaling Using Sigmoidal Contrast Enhancement Functions", http://markfairchild.org/PDFs/PAP07.pdf Args: image: Image to be adjusted in the shape of :math:`(*, H, W)`. cutoff: The cutoff of sigmoid function. gain: The multiplier of sigmoid function. inv: If is set to True the function will return the negative sigmoid correction. Example: >>> x = torch.ones(1, 1, 2, 2) >>> AdjustSigmoid(gain=0)(x) tensor([[[[0.5000, 0.5000], [0.5000, 0.5000]]]]) rdreFrfr)rLrgrYrrc t||_||_||_dSr)rrrfrLrg)rrfrLrgrr7r8rd  zAdjustSigmoid.__init__rr cCt||j|j|jdS)N)rfrLrg)rirfrLrgrrr7r7r8rjzAdjustSigmoid.forwardrdreF)rfr)rLr)rgrYrrrr rr rr7r7rr8rMsrcr) AdjustLogaAdjust log correction on the input image tensor. The input image is expected to be in the range of [0, 1]. Reference: [1]: http://www.ece.ucsb.edu/Faculty/Manjunath/courses/ece178W03/EnhancePart1.pdf Args: image: Image to be adjusted in the shape of :math:`(*, H, W)`. gain: The multiplier of logarithmic function. inv: If is set to True the function will return the inverse logarithmic correction. clip_output: Whether to clip the output image with range of [0, 1]. Example: >>> x = torch.zeros(1, 1, 2, 2) >>> AdjustLog(inv=True)(x) tensor([[[[0., 0.], [0., 0.]]]]) r#FTrLr)rgrYrXrrcrr)rrrLrgrX)rrLrgrXrr7r8rrzAdjustLog.__init__rr cCr)N)rLrgrX)rkrLrgrXrr7r7r8rrzAdjustLog.forwardr#FT)rLr)rgrYrXrYrrrrr7r7rr8rnsrcr)AdjustBrightnessAccumulativea{Adjust Brightness of an image accumulatively. This implementation aligns PIL. Hence, the output is close to TorchVision. The input image is expected to be in the range of [0, 1]. Args: brightness_factor: Brightness adjust factor per element in the batch. 0 does not modify the input image while any other number modify the brightness. Shape: - Input: Image/Input to be adjusted in the shape of :math:`(*, N)`. - Output: Adjusted image in the shape of :math:`(*, N)`. Example: >>> x = torch.ones(1, 1, 3, 3) >>> AdjustBrightnessAccumulative(1.)(x) tensor([[[[1., 1., 1.], [1., 1., 1.], [1., 1., 1.]]]]) >>> x = torch.ones(2, 5, 3, 3) >>> y = torch.ones(2) >>> AdjustBrightnessAccumulative(y)(x).shape torch.Size([2, 5, 3, 3]) r rrrcrrr rrr7r8rrz%AdjustBrightnessAccumulative.__init__rJr cCrr)rcr rr7r7r8rrz$AdjustBrightnessAccumulative.forwardrrrr7r7rr8rrrcs.eZdZdZd dfdd Zdd d ZZS)InvertaInvert the values of an input tensor by its maximum value. Args: input: The input tensor to invert with an arbitatry shape. max_val: The expected maximum value in the input tensor. The shape has to according to the input tensor shape, or at least has to work with broadcasting. Default: 1.0. Example: >>> img = torch.rand(1, 2, 4, 4) >>> Invert()(img).shape torch.Size([1, 2, 4, 4]) >>> img = 255. * torch.rand(1, 2, 3, 4, 4) >>> Invert(torch.as_tensor(255.))(img).shape torch.Size([1, 2, 3, 4, 4]) >>> img = torch.rand(1, 3, 4, 4) >>> Invert(torch.as_tensor([[[[1.]]]]))(img).shape torch.Size([1, 3, 4, 4]) Nrrrrcs@t|durtd}t|ts|d|dS||_dS)Nr;r)rrr*r r(r register_bufferr)rrrr7r8rs    zInvert.__init__rJr cCrr)rrrr7r7r8rrzInvert.forwardr)rrrrrrr7r7rr8rs r)rr rrrr r)rJr rKrrLrrr )T)rr rrrXrYrr r) rr rfr)rLr)rgrYrr r) rr rLr)rgrYrXrYrr )rd)rJr rlrrr )rdN)rJr rlrr~rrr )rJr rrrr )rJr rrrr )rr rr rr rr )rr rr rr )rr rr rr)rr rrrr )= __future__rmathrtypingrrrr* kornia.colorrrr kornia.corer Moduler r r kornia.core.checkrrrkornia.utils.helpersrkornia.utils.imagerrr9r>rBrHrIrWr\r`rbrcrirkr}rrrrrrrrrrrrrrr r rrrrr7r7r7r8sb       9 * ! + H D : D 5 " &! E X  7  *   %033'&'%! %