2Vh?AdZddlZddlZddlZddlZddlmZddlm Z  ddl m Z  e jZeEej"ej$ej&ej(ej*ej,dZe ddgdd Ze d d gdd Ze d dgddZe ddg ddZe d ddZy#e$re ZYwxYw#e$rdZ dZYwxYw)z$Utilities related to image handling.N)backend) keras_export)Image)nearestbilinearbicubichammingboxlanczoszkeras.utils.array_to_imgz&keras.preprocessing.image.array_to_imgctj|}|tj}t t dt j ||}|jdk7rtd|j|dk(r|jddd }|r<|t j|z }t j|}|d k7r||z}|d z}|jdd k(r%tj|jd d S|jddk(r%tj|jd dS|jddk(rvt j|d kDr/tj|ddddd fjddStj|ddddd fjd dStd|jd)aConverts a 3D NumPy array to a PIL Image instance. Example: ```python from PIL import Image img = np.random.random(size=(100, 100, 3)) pil_img = keras.utils.array_to_img(img) ``` Args: x: Input data, in any form that can be converted to a NumPy array. data_format: Image data format, can be either `"channels_first"` or `"channels_last"`. Defaults to `None`, in which case the global setting `keras.backend.image_data_format()` is used (unless you changed it, it defaults to `"channels_last"`). scale: Whether to rescale the image such that minimum and maximum values are 0 and 255 respectively. Defaults to `True`. dtype: Dtype to use. `None` means the global setting `keras.backend.floatx()` is used (unless you changed it, it defaults to `"float32"`). Defaults to `None`. Returns: A PIL Image instance. NzCCould not import PIL.Image. The use of `array_to_img` requires PIL.dtypezJExpected image array to have rank 3 (single image). Got array with shape: channels_firstruint8RGBARGBint32ILzUnsupported channel number: )rstandardize_data_formatfloatx pil_image ImportErrornpasarrayndim ValueErrorshape transposeminmax fromarrayastype)x data_formatscalerx_maxs K/home/dcms/DCMS/lib/python3.12/site-packages/keras/src/utils/image_utils.py array_to_imgr.#sB11+>K }  6   1E"Avv{ %%&WWI /  && KK1a  q Mq  A: JA SwwqzQ""188G#4f== q""188G#4e<< q 66!9s?&&qAqz'8'8'A3G G""1Q1W:#4#4W#=sCC7 |DEEzkeras.utils.img_to_arrayz&keras.preprocessing.image.img_to_arrayctj|}|tj}tj||}t |j dk(r|dk(r|jddd}|St |j dk(re|dk(r0|jd|j d|j df}|S|j|j d|j ddf}|Std|j )a(Converts a PIL Image instance to a NumPy array. Example: ```python from PIL import Image img_data = np.random.random(size=(100, 100, 3)) img = keras.utils.array_to_img(img_data) array = keras.utils.image.img_to_array(img) ``` Args: img: Input PIL Image instance. data_format: Image data format, can be either `"channels_first"` or `"channels_last"`. Defaults to `None`, in which case the global setting `keras.backend.image_data_format()` is used (unless you changed it, it defaults to `"channels_last"`). dtype: Dtype to use. `None` means the global setting `keras.backend.floatx()` is used (unless you changed it, it defaults to `"float32"`). Returns: A 3D NumPy array. r rrrrrzUnsupported image shape: ) rrrrr lenr#r$reshaper")imgr*rr)s r- img_to_arrayr4ns@11+>K }  3e$A 177|q * * Aq!$A H QWW  * * 1aggaj!''!*56A H 1771:qwwqz156A H4QWWI>??r/zkeras.utils.save_imgz"keras.preprocessing.image.save_imgc tj|}t|||}|jdk(r0|dk(s|dk(r&t j d|j d}|j|fd|i|y) apSaves an image stored as a NumPy array to a path or file object. Args: path: Path or file object. x: NumPy array. data_format: Image data format, either `"channels_first"` or `"channels_last"`. file_format: Optional file format override. If omitted, the format to use is determined from the filename extension. If a file object was used instead of a filename, this parameter should always be used. scale: Whether to rescale image values to be within `[0, 255]`. **kwargs: Additional keyword arguments passed to `PIL.Image.save()`. )r*r+rjpgjpegz?The JPG format does not support RGBA images, converting to RGB.rformatN)rrr.modewarningswarnconvertsave)pathr)r* file_formatr+kwargsr3s r-save_imgrAss11+>K qk ?C xx6{e3{f7L M kk%  CHHT0+00r/zkeras.utils.load_imgz"keras.preprocessing.image.load_imgc t tdt|tjrtj |}nt|t jttfrt|t jrt|j}t |d5}tj tj|j}dddntdt||dk(r jdvri|jd}nW|dk(r!jd k7rC|jd }n1|d k(r!jd k7r|jd }n t!d ||d |df}|j"|k7r|t$vr color_mode target_size interpolationkeep_aspect_ratior3fwidth_height_tupleresamplewidthheight target_width target_height crop_height crop_widthcrop_box_hstartcrop_box_wstart crop_box_wend crop_box_hendcrop_boxs r-load_imgrjsN M  $ #nnT" D7<<4 5 dGLL )t||~&D $  7..AFFH!56C 7 7:4:, G  [  88- -++c"C v  88v ++f%C u  88u ++e$CKLL)!nk!n= 88) )$== %%+V% ";"@"@"BC&1?H # v.@+ m$}4E $|3 E "&+6   3 #)K#7A"=#(:#5!"; /* < /+ = ##!!  jj!3X8jL Jjj!3X> Ju 7 7s &7I))I3z&keras.preprocessing.image.smart_resizec|xst}t|dk7rtd|d|j|}t|jIt|jdkst|jdkDrtd|jd|j |}|dk(r |d |d }}n |d |d }}|\} } t |t rt |t rt t|| z| z } tt|| d } t t|| z| z } tt|| d } t t|| z dz } t t|| z dz }n4|j|j|| zd | z d} |jj|| } |jj| d } |j| d} |j|j|| zd | z d} |jj|| } |jj| d } |j| d} |j|j|| z d dz d} |j|j|| z d dz d}|dk(rEt|jdk(r|dd| | | z||| zddf}nY|| | | z||| zddf}nDt|jdk(r|dddd| | | z||| zf}n|dd| | | z||| zf}|jj||||}t |t j"rt!j$|S|S)a Resize images to a target size without aspect ratio distortion. Image datasets typically yield images that have each a different size. However, these images need to be batched before they can be processed by Keras layers. To be batched, images need to share the same height and width. You could simply do, in TF (or JAX equivalent): ```python size = (200, 200) ds = ds.map(lambda img: resize(img, size)) ``` However, if you do this, you distort the aspect ratio of your images, since in general they do not all have the same aspect ratio as `size`. This is fine in many cases, but not always (e.g. for image generation models this can be a problem). Note that passing the argument `preserve_aspect_ratio=True` to `resize` will preserve the aspect ratio, but at the cost of no longer respecting the provided target size. This calls for: ```python size = (200, 200) ds = ds.map(lambda img: smart_resize(img, size)) ``` Your output images will actually be `(200, 200)`, and will not be distorted. Instead, the parts of the image that do not fit within the target size get cropped out. The resizing process is: 1. Take the largest centered crop of the image that has the same aspect ratio as the target size. For instance, if `size=(200, 200)` and the input image has size `(340, 500)`, we take a crop of `(340, 340)` centered along the width. 2. Resize the cropped image to the target size. In the example above, we resize the `(340, 340)` crop to `(200, 200)`. Args: x: Input image or batch of images (as a tensor or NumPy array). Must be in format `(height, width, channels)` or `(batch_size, height, width, channels)`. size: Tuple of `(height, width)` integer. Target size. interpolation: String, interpolation to use for resizing. Supports `"bilinear"`, `"nearest"`, `"bicubic"`, `"lanczos3"`, `"lanczos5"`. Defaults to `"bilinear"`. data_format: `"channels_last"` or `"channels_first"`. backend_module: Backend module to use (if different from the default backend). Returns: Array with shape `(size[0], size[1], channels)`. If the input image was a NumPy array, the output is a NumPy array, and if it was a backend-native tensor, the output is a backend-native tensor. rz6Expected `size` to be a tuple of 2 integers, but got: .NrrzExpected an image array with shape `(height, width, channels)`, or `(batch_size, height, width, channels)`, but got input with incorrect rank, of shape channels_lastrfloat32r)rSrZr*)rr1r"convert_to_tensorr#rGintfloatr&r%castnumpyminimummaximumimagerWrndarrayarray)r)rSrZr*backend_moduler3r#r`r_rbrarcrdrerfs r- smart_resizer}(sL$.wN 4yA~DTF! L    * *1 -C 399~! syy>A SYY!!3;;>99+QH    %Eo%b 59b 59"&M<&#:eS#9% 56EF #fk2A6 v 45 EF UJ/3 eF[$89A=>eEJ$67!;<$))    5y A   %**226;G $**22;B $))+w? #((    5y A   $))11%D #))11*a@ #((W= (--    4i @1 Dg )--    2I > BG o% syy>Q /K"??/J">>C/K"??/J">>C syy>Q /K"??/J">>@C/K"??/J">>@C    % % $m & C!RZZ xx} Jr/)NTN)NN)NNT)rFNrF)rrmN)__doc__rHrKr:rvr keras.srcrkeras.src.api_exportrPILrr Resamplingpil_image_resamplingAttributeErrorrNEARESTBILINEARBICUBICHAMMINGBOXLANCZOSrTr.r4rArjr}r/r-rsr* -  &)(33#'//(11'//'//#'''// !"0 BF  BFJ"0 +  + \%'KLM1N10%'KLM iNiX67 `8`q )() I s.C" CCC"CC"" C.-C.