o i er @s@dZddlZddlmmZddlmZddl m Z ddl m Z ddl m Z ddlmZddlmZdd lmZd Zd Zd d ZeddGddde jZeddGddde jZeddgdGddde jZeddGddde jZdZd Zd!Zed"d#gdGd$d%d%e jZ ed&d'gdGd(d)d)e jZ!dNd*d+Z" , - .  dOd/d0Z#dNd1d2Z$ed3d4gdGd5d6d6e jZ%ed7d8gdGd9d:d:e jZ&dNd;d<Z'ed=d>gdGd?d@d@e jZ(edAgdGdBdCdCe jZ)edDdEgdGdFdGdGe jZ*edHdIgdGdJdKdKe jZ+dNdLdMZ,dS)Pz!Keras image preprocessing layers.N) keras_export)backend) base_layer)base_preprocessing_layer)preprocessing_utils) image_utils)tf_utilscCs4|dvr td|d|dvrtd|ddS)N>wrapnearestreflectconstantzUnknown `fill_mode` zA. Only `reflect`, `wrap`, `constant` and `nearest` are supported.>r bilinearzUnknown `interpolation` z.. Only `nearest` and `bilinear` are supported.)NotImplementedError) fill_mode interpolationra/var/www/myenv/lib/python3.10/site-packages/keras/src/layers/preprocessing/image_preprocessing.py!check_fill_mode_and_interpolation s  rzkeras.layers.Resizingz0keras.layers.experimental.preprocessing.ResizingcsBeZdZdZ  d fdd ZddZdd Zfd d ZZS) ResizingaA preprocessing layer which resizes images. This layer resizes an image input to a target height and width. The input should be a 4D (batched) or 3D (unbatched) tensor in `"channels_last"` format. Input pixel values can be of any range (e.g. `[0., 1.)` or `[0, 255]`) and of integer or floating point dtype. By default, the layer will output floats. This layer can be called on tf.RaggedTensor batches of input images of distinct sizes, and will resize the outputs to dense tensors of uniform size. For an overview and full list of preprocessing layers, see the preprocessing [guide](https://www.tensorflow.org/guide/keras/preprocessing_layers). Args: height: Integer, the height of the output shape. width: Integer, the width of the output shape. interpolation: String, the interpolation method. Supports `"bilinear"`, `"nearest"`, `"bicubic"`, `"area"`, `"lanczos3"`, `"lanczos5"`, `"gaussian"`, `"mitchellcubic"`. Defaults to `"bilinear"`. crop_to_aspect_ratio: If True, resize the images without aspect ratio distortion. When the original aspect ratio differs from the target aspect ratio, the output image will be cropped so as to return the largest possible window in the image (of size `(height, width)`) that matches the target aspect ratio. By default (`crop_to_aspect_ratio=False`), aspect ratio may not be preserved. rFc sL||_||_||_||_t||_tjdi|t j d ddS)NrTr) heightwidthrcrop_to_aspect_ratiorget_interpolation_interpolation_methodsuper__init__rkeras_kpl_gaugeget_cellset)selfrrrrkwargs __class__rrrPszResizing.__init__csjdkr j}ntj}t|djjgjrJfdd}t rEt }|j dd}t ||}tj ||d}n|}n tjjjd}t|jS)Nr dtypecs$tr |}tj|jdS)N)sizer)r is_ragged to_tensorr smart_resizer)xinputsr!r'rrresize_to_aspectos z'Resizing.call..resize_to_aspect)fn_output_signature)r'method)r compute_dtypetffloat32convert_inputsrrrrr( TensorShapeshape TensorSpecmap_fnimageresizercast)r!r- input_dtyper. size_as_shaper7specoutputsrr,rcallbs&       z Resizing.callcC,t|}|j|t<|j|t<t|SNr3r6as_listrH_AXISrW_AXISr! input_shaperrrcompute_output_shape   zResizing.compute_output_shapecs<|j|j|j|jd}t}tt|t|S)N)rrrr) rrrrr get_configdictlistitemsr!config base_configr#rrrLs zResizing.get_config)rF __name__ __module__ __qualname____doc__rrArJrL __classcell__rrr#rr-s##rzkeras.layers.CenterCropz2keras.layers.experimental.preprocessing.CenterCropcs<eZdZdZfddZddZddZfdd ZZS) CenterCropaA preprocessing layer which crops images. This layers crops the central portion of the images to a target size. If an image is smaller than the target size, it will be resized and cropped so as to return the largest possible window in the image that matches the target aspect ratio. Input pixel values can be of any range (e.g. `[0., 1.)` or `[0, 255]`) and of integer or floating point dtype. By default, the layer will output floats. For an overview and full list of preprocessing layers, see the preprocessing [guide](https://www.tensorflow.org/guide/keras/preprocessing_layers). Input shape: 3D (unbatched) or 4D (batched) tensor with shape: `(..., height, width, channels)`, in `"channels_last"` format. Output shape: 3D (unbatched) or 4D (batched) tensor with shape: `(..., target_height, target_width, channels)`. If the input height/width is even and the target height/width is odd (or inversely), the input image is left-padded by 1 pixel. Args: height: Integer, the height of the output shape. width: Integer, the width of the output shape. c s<||_||_tjdi|dditjdddS)NautocastFrYTr)rrrrrrrr )r!rrr"r#rrrs  zCenterCrop.__init__csrtjt}|tj|tjfdd}fdd}tt dkdkf||S)Ncs<tdtj}tdtj}tj||jjS)N)r3r<int32r:crop_to_bounding_boxrr)h_startw_starth_diffr-r!w_diffrr center_crops z$CenterCrop.call..center_cropc"tjjg}t|jSrCrr*rrr3r<r2r@r-r!rrupsize zCenterCrop.call..upsizer r5r2r3r7rFrrGrcond reduce_all)r!r-rIrcrhrr`rrAs  zCenterCrop.callcCrBrCrDrHrrrrJrKzCenterCrop.compute_output_shapec4|j|jd}t}tt|t|S)N)rr)rrrrLrMrNrOrPr#rrrL  zCenterCrop.get_configrSrrr#rrYs  rYzkeras.layers.RandomCropz2keras.layers.experimental.preprocessing.RandomCrop)v1c@eZdZdZd fdd Zd ddZdd Zfd d ZZS) RandomCropaoA preprocessing layer which randomly crops images during training. During training, this layer will randomly choose a location to crop images down to a target size. The layer will crop all the images in the same batch to the same cropping location. At inference time, and during training if an input image is smaller than the target size, the input will be resized and cropped so as to return the largest possible window in the image that matches the target aspect ratio. If you need to apply random cropping at inference time, set `training` to True when calling the layer. Input pixel values can be of any range (e.g. `[0., 1.)` or `[0, 255]`) and of integer or floating point dtype. By default, the layer will output floats. For an overview and full list of preprocessing layers, see the preprocessing [guide](https://www.tensorflow.org/guide/keras/preprocessing_layers). Input shape: 3D (unbatched) or 4D (batched) tensor with shape: `(..., height, width, channels)`, in `"channels_last"` format. Output shape: 3D (unbatched) or 4D (batched) tensor with shape: `(..., target_height, target_width, channels)`. Args: height: Integer, the height of the output shape. width: Integer, the width of the output shape. seed: Integer. Used to create a random seed. Nc sFtjddtjdi|d|dd||_||_||_dS)NrqTF)rZseedforce_generatorr) rrrr rrrrrr)r!rrrrr"r#rrrs    zRandomCrop.__init__Tcsxtjdttjtjfdd}fdd}tt |dkdkf||S)Nr%csTj}jdgd|j|}|dd}|dd}tj||jjS)Nr[r) r&_random_generatorrandom_uniformmaxr3r:r]rr)r&randsr^r_rarIr-r!rbrr random_crop s z$RandomCrop.call..random_cropcrdrCrerfrgrrr;+rizRandomCrop.call..resizerrj)r!r-trainingrzr;rryrrAs  zRandomCrop.callcCrBrCrDrHrrrrJ8rKzRandomCrop.compute_output_shapec8|j|j|jd}t}tt|t|S)N)rrrr)rrrrrrLrMrNrOrPr#rrrL>  zRandomCrop.get_configrCTrSrrr#rrqs ! rqzkeras.layers.Rescalingz1keras.layers.experimental.preprocessing.Rescalingcs>eZdZdZd fdd ZddZddZfd d ZZS) RescalingaA preprocessing layer which rescales input values to a new range. This layer rescales every value of an input (often an image) by multiplying by `scale` and adding `offset`. For instance: 1. To rescale an input in the `[0, 255]` range to be in the `[0, 1]` range, you would pass `scale=1./255`. 2. To rescale an input in the `[0, 255]` range to be in the `[-1, 1]` range, you would pass `scale=1./127.5, offset=-1`. The rescaling is applied both during training and inference. Inputs can be of integer or floating point dtype, and by default the layer will output floats. For an overview and full list of preprocessing layers, see the preprocessing [guide](https://www.tensorflow.org/guide/keras/preprocessing_layers). Input shape: Arbitrary. Output shape: Same as input. Args: scale: Float, the scale to apply to the inputs. offset: Float, the offset to apply to the inputs. c s4||_||_tjdi|tjdddS)NrTr)scaleoffsetrrrrrr )r!rrr"r#rrrlszRescaling.__init__cCsB|j}t||d}t|j|}t|j|}t||||SNr%)r2r5r3r<rr)r!r-r&rrrrrrArs  zRescaling.callcC|SrCrrHrrrrJyzRescaling.compute_output_shapecrm)N)rr)rrrrLrMrNrOrPr#rrrL|rnzRescaling.get_config)rrSrrr#rrHs r horizontalverticalhorizontal_and_verticalzkeras.layers.RandomFlipz2keras.layers.experimental.preprocessing.RandomFlipcsDeZdZdZedffdd Zd ddZdd Zfd d ZZ S) RandomFlipaA preprocessing layer which randomly flips images during training. This layer will flip the images horizontally and or vertically based on the `mode` attribute. During inference time, the output will be identical to input. Call the layer with `training=True` to flip the input. Input pixel values can be of any range (e.g. `[0., 1.)` or `[0, 255]`) and of integer or floating point dtype. By default, the layer will output floats. For an overview and full list of preprocessing layers, see the preprocessing [guide](https://www.tensorflow.org/guide/keras/preprocessing_layers). Input shape: 3D (unbatched) or 4D (batched) tensor with shape: `(..., height, width, channels)`, in `"channels_last"` format. Output shape: 3D (unbatched) or 4D (batched) tensor with shape: `(..., height, width, channels)`, in `"channels_last"` format. Args: mode: String indicating which flip mode to use. Can be `"horizontal"`, `"vertical"`, or `"horizontal_and_vertical"`. `"horizontal"` is a left-right flip and `"vertical"` is a top-bottom flip. Defaults to `"horizontal_and_vertical"` seed: Integer. Used to create a random seed. Nc stjd|dd|tjdd||_|tkr#d|_d|_ n!|t kr.d|_d|_ n|t kr9d|_d|_ n t d|j d|||_dS)NTrrrsrFzRandomFlip layer z# received an unknown mode argument r)rrrrrr mode HORIZONTALrrVERTICALHORIZONTAL_AND_VERTICAL ValueErrornamerr)r!rrrr"r#rrrs(   zRandomFlip.__init__Tc(t|j}fdd}|r||S|S)Ncs|}jr!j}|durtjj||d}n tj|j}jr@j}|dur6tjj ||d}n tj |j}| |j |S)Nrr) rrumake_seed_for_stateless_opr3r: stateless_random_flip_left_rightrandom_flip_left_rightmake_legacy_seedrstateless_random_flip_up_downrandom_flip_up_down set_shaper7)r-flipped_outputsrrr!rrrandom_flipped_inputss.   z.RandomFlip.call..random_flipped_inputsr5r2)r!r-r{rrrrrAs  zRandomFlip.callcCrrCrrHrrrrJrzRandomFlip.compute_output_shapecrm)N)rrr)rrrrrLrMrNrOrPr#rrrLrnzRandomFlip.get_configr~) rTrUrVrWrrrArJrLrXrrr#rrs  #rzkeras.layers.RandomTranslationz9keras.layers.experimental.preprocessing.RandomTranslationcHeZdZdZ    dfdd Zdd d Zd d Zfd dZZS)RandomTranslationa A preprocessing layer which randomly translates images during training. This layer will apply random translations to each image during training, filling empty space according to `fill_mode`. Input pixel values can be of any range (e.g. `[0., 1.)` or `[0, 255]`) and of integer or floating point dtype. By default, the layer will output floats. For an overview and full list of preprocessing layers, see the preprocessing [guide](https://www.tensorflow.org/guide/keras/preprocessing_layers). Args: height_factor: a float represented as fraction of value, or a tuple of size 2 representing lower and upper bound for shifting vertically. A negative value means shifting image up, while a positive value means shifting image down. When represented as a single positive float, this value is used for both the upper and lower bound. For instance, `height_factor=(-0.2, 0.3)` results in an output shifted by a random amount in the range `[-20%, +30%]`. `height_factor=0.2` results in an output height shifted by a random amount in the range `[-20%, +20%]`. width_factor: a float represented as fraction of value, or a tuple of size 2 representing lower and upper bound for shifting horizontally. A negative value means shifting image left, while a positive value means shifting image right. When represented as a single positive float, this value is used for both the upper and lower bound. For instance, `width_factor=(-0.2, 0.3)` results in an output shifted left by 20%, and shifted right by 30%. `width_factor=0.2` results in an output height shifted left or right by 20%. fill_mode: Points outside the boundaries of the input are filled according to the given mode (one of `{"constant", "reflect", "wrap", "nearest"}`). - *reflect*: `(d c b a | a b c d | d c b a)` The input is extended by reflecting about the edge of the last pixel. - *constant*: `(k k k k | a b c d | k k k k)` The input is extended by filling all values beyond the edge with the same constant value k = 0. - *wrap*: `(a b c d | a b c d | a b c d)` The input is extended by wrapping around to the opposite edge. - *nearest*: `(a a a a | a b c d | d d d d)` The input is extended by the nearest pixel. interpolation: Interpolation mode. Supported values: `"nearest"`, `"bilinear"`. seed: Integer. Used to create a random seed. fill_value: a float represents the value to be filled outside the boundaries when `fill_mode="constant"`. Input shape: 3D (unbatched) or 4D (batched) tensor with shape: `(..., height, width, channels)`, in `"channels_last"` format. Output shape: 3D (unbatched) or 4D (batched) tensor with shape: `(..., height, width, channels)`, in `"channels_last"` format. r rNrc sHtjddtjd |dd|||_t|tt fr*|d|_ |d|_ n| |_ ||_ |j |j kr>t d|t |j dksLt |j dkrSt d|||_t|tt frh|d|_|d|_n| |_||_|j|jkr|t d |t |jdkst |jdkrt d |t||||_||_||_||_dS) NrTrrrtzC`height_factor` cannot have upper bound less than lower bound, got ?S`height_factor` argument must have values between [-1, 1]. Received: height_factor=zB`width_factor` cannot have upper bound less than lower bound, got z5`width_factor` must have values between [-1, 1], got r)rrrr rr height_factor isinstancetuplerN height_lower height_upperrabs width_factor width_lower width_upperrr fill_valuerrrr!rrrrrrrr"r#rrr1s\         zRandomTranslation.__init__Tcr)Nc s|j}|jjdk}|rt|d}t|}|d}t|ttj}t|ttj}jj |dgj j tjd}||}jj |dgj j tjd}||}tjtj||gddtjd} t|t| jjjd} |rwt| d} | || S)z"Translated inputs with random ops.rrt)r7minvalmaxvalr&axisr%)rrr)r7rankr3 expand_dimsr<rFr4rGrurvrrrrconcat transformget_translation_matrixrrrsqueezer) r-original_shape unbatched inputs_shape batch_sizeimg_hdimg_wdheight_translatewidth_translate translationsoutputrrrrandom_translated_inputsmsJ     z8RandomTranslation.call..random_translated_inputsr)r!r-r{rrrrrAjs  ,zRandomTranslation.callcCrrCrrHrrrrJrz&RandomTranslation.compute_output_shapecD|j|j|j|j|j|jd}t}tt | t | SN)rrrrrrr rrrrrrrrrLrMrNrOrPr#rrrL zRandomTranslation.get_configr rNrr~rSrrr#rrs< 94rc Cst|pdPt|d}tjt|dftjt|dftj|ddddf t|dftjt|dftj|ddddf t|dftjgddWdS1sZwYdS)ayReturns projective transform(s) for the given translation(s). Args: translations: A matrix of 2-element lists representing `[dx, dy]` to translate for each image (for a batch of images). name: The name of the op. Returns: A tensor of shape `(num_images, 8)` projective transforms which can be given to `transform`. translation_matrixrrtNr[valuesr)r name_scoper3r7ronesr4zeros)rrnum_translationsrrrrs  $rr rrc Cst|pdU|dur$t|dd}ts$t|}|dur$|}tj|tjdd}| dgst d|t ||||_||_||_||_dS)NrTrrrtzA`factor` argument cannot have a negative value. Received: factor=r)rrrr rrfactorrrrNlowerrrrrrrrr)r!rrrrrrr"r#rrrs*      zRandomRotation.__init__Tcr)Nc s|j}|jjdk}|rt|d}t|}|d}t|ttj}t|ttj}jdt j }j dt j }j j |g||d} t|t| ||jjjd} |r^t| d} | || S)zRotated inputs with random ops.rrrr7rrrrr)r7rr3rr<rFr4rGrnppirrurvrrrrrrr) r-rrrrrr min_angle max_anglerrrrrrandom_rotated_inputss0      z2RandomRotation.call..random_rotated_inputsr)r!r-r{rrrrrAs  zRandomRotation.callcCrrCrrHrrrrJrz#RandomRotation.compute_output_shapecs@|j|j|j|j|jd}t}tt| t| S)N)rrrrrr) rrrrrrrrLrMrNrOrPr#rrrLs zRandomRotation.get_configrr~rSrrr#rrfs> &rzkeras.layers.RandomZoomz2keras.layers.experimental.preprocessing.RandomZoomcsJeZdZdZ     dfdd Zdd d Zd d Zfd dZZS) RandomZoomaX A preprocessing layer which randomly zooms images during training. This layer will randomly zoom in or out on each axis of an image independently, filling empty space according to `fill_mode`. Input pixel values can be of any range (e.g. `[0., 1.)` or `[0, 255]`) and of integer or floating point dtype. By default, the layer will output floats. For an overview and full list of preprocessing layers, see the preprocessing [guide](https://www.tensorflow.org/guide/keras/preprocessing_layers). Args: height_factor: a float represented as fraction of value, or a tuple of size 2 representing lower and upper bound for zooming vertically. When represented as a single float, this value is used for both the upper and lower bound. A positive value means zooming out, while a negative value means zooming in. For instance, `height_factor=(0.2, 0.3)` result in an output zoomed out by a random amount in the range `[+20%, +30%]`. `height_factor=(-0.3, -0.2)` result in an output zoomed in by a random amount in the range `[+20%, +30%]`. width_factor: a float represented as fraction of value, or a tuple of size 2 representing lower and upper bound for zooming horizontally. When represented as a single float, this value is used for both the upper and lower bound. For instance, `width_factor=(0.2, 0.3)` result in an output zooming out between 20% to 30%. `width_factor=(-0.3, -0.2)` result in an output zooming in between 20% to 30%. `None` means i.e., zooming vertical and horizontal directions by preserving the aspect ratio. Defaults to `None`. fill_mode: Points outside the boundaries of the input are filled according to the given mode (one of `{"constant", "reflect", "wrap", "nearest"}`). - *reflect*: `(d c b a | a b c d | d c b a)` The input is extended by reflecting about the edge of the last pixel. - *constant*: `(k k k k | a b c d | k k k k)` The input is extended by filling all values beyond the edge with the same constant value k = 0. - *wrap*: `(a b c d | a b c d | a b c d)` The input is extended by wrapping around to the opposite edge. - *nearest*: `(a a a a | a b c d | d d d d)` The input is extended by the nearest pixel. interpolation: Interpolation mode. Supported values: `"nearest"`, `"bilinear"`. seed: Integer. Used to create a random seed. fill_value: a float represents the value to be filled outside the boundaries when `fill_mode="constant"`. Example: >>> input_img = np.random.random((32, 224, 224, 3)) >>> layer = tf.keras.layers.RandomZoom(.5, .2) >>> out_img = layer(input_img) >>> out_img.shape TensorShape([32, 224, 224, 3]) Input shape: 3D (unbatched) or 4D (batched) tensor with shape: `(..., height, width, channels)`, in `"channels_last"` format. Output shape: 3D (unbatched) or 4D (batched) tensor with shape: `(..., height, width, channels)`, in `"channels_last"` format. Nr rrc stjddtjd |dd|||_t|tt fr*|d|_ |d|_ n| |_ ||_ t |j dks?t |j dkrFt d|||_|durwt|tt fr_|d|_|d|_n| |_||_|jdksp|jdkrwt d |t||||_||_||_||_dS) NrTrrrtrrzP`width_factor` argument must have values larger than -1. Received: width_factor=r)rrrr rrrrrrNrrrrrrrrrrrrrrr#rrrIsD       zRandomZoom.__init__Tcr)Nc s|j}|jjdk}|rt|d}t|}|d}t|ttj}t|ttj}jj |dgdj dj d}j durTjj |dgdj djd}n|}tjtj||gddtjd} t|t| ||jjjd } |r}t| d} | || S) zZoomed inputs with random ops.rrrtrrNrr%r)r7rr3rr<rFr4rGrurvrrrrrrrget_zoom_matrixrrrrr) r-rrrrrr height_zoom width_zoomzoomsrrrrrandom_zoomed_inputs~sD       z-RandomZoom.call..random_zoomed_inputsr)r!r-r{rrrrrA{s  )zRandomZoom.callcCrrCrrHrrrrJrzRandomZoom.compute_output_shapecrrrrPr#rrrLrzRandomZoom.get_config)Nr rNrr~rSrrr#rrsK 21rc Cst|pdbt|d}|ddd|ddddf}|ddd|ddddf}tj|ddddft|dftj|t|dftj|ddddf|t|dftjgddWdS1slwYdS) a#Returns projective transform(s) for the given zoom(s). Args: zooms: A matrix of 2-element lists representing `[zx, zy]` to zoom for each image (for a batch of images). image_height: Height of the image(s) to be transformed. image_width: Width of the image(s) to be transformed. name: The name of the op. Returns: A tensor of shape `(num_images, 8)`. Projective transforms which can be given to operation `image_projective_transform_v2`. If one row of transforms is `[a0, a1, a2, b0, b1, b2, c0, c1]`, then it maps the *output* point `(x, y)` to a transformed *input* point `(x', y') = ((a0 x + a1 y + a2) / k, (b0 x + b1 y + b2) / k)`, where `k = c0 x + c1 y + 1`. zoom_matrixrrrNrtr[r)rrr3r7rrr4)rrrr num_zoomsrrrrrrs "" $rzkeras.layers.RandomContrastz6keras.layers.experimental.preprocessing.RandomContrastcrp)RandomContrastaA preprocessing layer which randomly adjusts contrast during training. This layer will randomly adjust the contrast of an image or images by a random factor. Contrast is adjusted independently for each channel of each image during training. For each channel, this layer computes the mean of the image pixels in the channel and then adjusts each component `x` of each pixel to `(x - mean) * contrast_factor + mean`. Input pixel values can be of any range (e.g. `[0., 1.)` or `[0, 255]`) and in integer or floating point dtype. By default, the layer will output floats. The output value will be clipped to the range `[0, 255]`, the valid range of RGB colors. For an overview and full list of preprocessing layers, see the preprocessing [guide](https://www.tensorflow.org/guide/keras/preprocessing_layers). Input shape: 3D (unbatched) or 4D (batched) tensor with shape: `(..., height, width, channels)`, in `"channels_last"` format. Output shape: 3D (unbatched) or 4D (batched) tensor with shape: `(..., height, width, channels)`, in `"channels_last"` format. Args: factor: a positive float represented as fraction of value, or a tuple of size 2 representing lower and upper bound. When represented as a single float, lower = upper. The contrast factor will be randomly picked between `[1.0 - lower, 1.0 + upper]`. For any pixel x in the channel, the output will be `(x - mean) * factor + mean` where `mean` is the mean value of the channel. seed: Integer. Used to create a random seed. Nc stjddtjd |dd|||_t|tt fr*|d|_ |d|_ n||_ |_ |j dks?|j dks?|j dkrFt d|||_ dS) NrTrrrtrrzX`factor` argument cannot have negative values or values greater than 1.Received: factor=r)rrrr rrrrrrNrrrrr)r!rrrr"r#rrrs     zRandomContrast.__init__Tcr)Ncsxj}|durtjj|djdj|d}ntjj|djdjjd}t |dd}| |j |S)Nrrr) rurr3r:stateless_random_contrastrrrandom_contrastr clip_by_valuerr7)r-rrrrrrrandom_contrasted_inputs)s  z5RandomContrast.call..random_contrasted_inputsr)r!r-r{rrrrrA&s  zRandomContrast.callcCrrCrrHrrrrJ?rz#RandomContrast.compute_output_shapecrm)N)rrr)rrrrrLrMrNrOrPr#rrrLBrnzRandomContrast.get_configrCr~rSrrr#rrs & rzkeras.layers.RandomBrightnesscs`eZdZdZdZdZdfdd Zdd Zd d Zd d Z dddZ ddZ fddZ Z S)RandomBrightnessap A preprocessing layer which randomly adjusts brightness during training. This layer will randomly increase/reduce the brightness for the input RGB images. At inference time, the output will be identical to the input. Call the layer with `training=True` to adjust the brightness of the input. Note that different brightness adjustment factors will be apply to each the images in the batch. For an overview and full list of preprocessing layers, see the preprocessing [guide](https://www.tensorflow.org/guide/keras/preprocessing_layers). Args: factor: Float or a list/tuple of 2 floats between -1.0 and 1.0. The factor is used to determine the lower bound and upper bound of the brightness adjustment. A float value will be chosen randomly between the limits. When -1.0 is chosen, the output image will be black, and when 1.0 is chosen, the image will be fully white. When only one float is provided, eg, 0.2, then -0.2 will be used for lower bound and 0.2 will be used for upper bound. value_range: Optional list/tuple of 2 floats for the lower and upper limit of the values of the input data. To make no change, use [0.0, 1.0], e.g., if the image input has been scaled before this layer. Defaults to [0.0, 255.0]. The brightness adjustment will be scaled to this range, and the output values will be clipped to this range. seed: optional integer, for fixed RNG behavior. Inputs: 3D (HWC) or 4D (NHWC) tensor, with float or int dtype. Input pixel values can be of any range (e.g. `[0., 1.)` or `[0, 255]`) Output: 3D (HWC) or 4D (NHWC) tensor with brightness adjusted based on the `factor`. By default, the layer will output floats. The output value will be clipped to the range `[0, 255]`, the valid range of RGB colors, and rescaled based on the `value_range` if needed. Sample usage: ```python random_bright = tf.keras.layers.RandomBrightness(factor=0.2) # An image with shape [2, 2, 3] image = [[[1, 2, 3], [4 ,5 ,6]], [[7, 8, 9], [10, 11, 12]]] # Assume we randomly select the factor to be 0.1, then it will apply # 0.1 * 255 to all the channel output = random_bright(image, training=True) # output will be int64 with 25.5 added to each channel and round down. tf.Tensor([[[26.5, 27.5, 28.5] [29.5, 30.5, 31.5]] [[32.5, 33.5, 34.5] [35.5, 36.5, 37.5]]], shape=(2, 2, 3), dtype=int64) ``` z^The `factor` argument should be a number (or a list of two numbers) in the range [-1.0, 1.0]. zt d||j dksH|j dkrOt d|||_ t||_||_dS) NrTrrrtz[`factor` argument cannot have an upper bound lesser than the lower bound. Received: factor=rD`factor` argument must have values larger than -1. Received: factor=r)rrrr rrrrrrNrrrrrrrrrr!rrrrr"r#rrrs6     zRandomHeight.__init__Tc$t|}fdd}|r||S|S)Nc st|}t|ttj}|t}jjgdjdj d}t||tj }t ||g}tj j ||jd}t|j}|j}d|t<|||S)z'Inputs height-adjusted with random ops.rrrr'r1N)r3r7r<rFr4rGrurvrrr\stackr:r;rr2rEr) r-rrrradjusted_height adjusted_sizerrrrrrandom_height_inputs4s(   z/RandomHeight.call..random_height_inputsr5)r!r-r{rrrrrA1  zRandomHeight.callcC t|}d|t<t|SrC)r3r6rErFrHrrrrJQ z!RandomHeight.compute_output_shapecr|N)rrrrrrrrrrLrMrNrOrPr#rrrLVr}zRandomHeight.get_configrNr~rSrrr#rrs )  rzkeras.layers.RandomWidthz3keras.layers.experimental.preprocessing.RandomWidthcr) RandomWidthaA preprocessing layer which randomly varies image width during training. This layer will randomly adjusts the width of a batch of images of a batch of images by a random factor. The input should be a 3D (unbatched) or 4D (batched) tensor in the `"channels_last"` image data format. Input pixel values can be of any range (e.g. `[0., 1.)` or `[0, 255]`) and of integer or floating point dtype. By default, the layer will output floats. By default, this layer is inactive during inference. For an overview and full list of preprocessing layers, see the preprocessing [guide](https://www.tensorflow.org/guide/keras/preprocessing_layers). Args: factor: A positive float (fraction of original width), or a tuple of size 2 representing lower and upper bound for resizing horizontally. When represented as a single float, this value is used for both the upper and lower bound. For instance, `factor=(0.2, 0.3)` results in an output with width changed by a random amount in the range `[20%, 30%]`. `factor=(-0.2, 0.3)` results in an output with width changed by a random amount in the range `[-20%, +30%]`. `factor=0.2` results in an output with width changed by a random amount in the range `[-20%, +20%]`. interpolation: String, the interpolation method. Supports `"bilinear"`, `"nearest"`, `"bicubic"`, `"area"`, `"lanczos3"`, `"lanczos5"`, `"gaussian"`, `"mitchellcubic"`. Defaults to `bilinear`. seed: Integer. Used to create a random seed. Input shape: 3D (unbatched) or 4D (batched) tensor with shape: `(..., height, width, channels)`, in `"channels_last"` format. Output shape: 3D (unbatched) or 4D (batched) tensor with shape: `(..., height, random_width, channels)`. rNc r) Nr&TrrrtzY`factor` argument cannot have an upper bound less than the lower bound. Received: factor=rrr)rrrr rrrrrrNrrrrrrrrrrr#rrrs6     zRandomWidth.__init__Tcr)Nc st|}|t}t|ttj}jjgdjdj d}t||tj }t ||g}tj j ||jd}t|j}|j}d|t<|||S)z&Inputs width-adjusted with random ops.rrrN)r3r7rFr<rGr4rurvrrr\rr:r;rr2rEr) r-rrrradjusted_widthrrrrrrrandom_width_inputss(   z-RandomWidth.call..random_width_inputsr)r!r-r{r(rrrrAr zRandomWidth.callcCr!rC)r3r6rErGrHrrrrJr"z RandomWidth.compute_output_shapecr|r#r$rPr#rrrLr}zRandomWidth.get_configr%r~rSrrr#rr&`s (  r&cCs0t|trtdt|dtj||d}|S)NzhThis layer can only process a tensor representing an image or a batch of images. Received: type(inputs)=z.If you need to pass a dict containing images, labels, and bounding boxes, you should instead use the preprocessing and augmentation layers from `keras_cv.layers`. See docs at https://keras.io/api/keras_cv/layers/r%)rrMrtypeutils ensure_tensor)r-r&rrrr5s  r5rC)r rrNN)-rWnumpyrtensorflow.compat.v2compatv2r3 tensorflow.python.util.tf_exportr keras.srcrkeras.src.enginerrkeras.src.layers.preprocessingrr*keras.src.utilsrrrFrGrLayerrrYBaseRandomLayerrqrrrrrrrrrrrrrrrr&r5rrrrs        fN[9d 7% b4 =+ _vt