~WhA&dZddlZddlZddlZddlmZddlmZddlm Z ddl m Z da dZ dZd Zd Zd Zd Zd ZdZe dddZeedee_e dddZdZdZ ddZdZdZdS)z@Utilities for ImageNet data preprocessing & prediction decoding.N) activations)backend) data_utils) keras_exportzUhttps://storage.googleapis.com/download.tensorflow.org/data/imagenet_class_index.jsonat Preprocesses a tensor or Numpy array encoding a batch of images. Usage example with `applications.MobileNet`: ```python i = tf.keras.layers.Input([None, None, 3], dtype = tf.uint8) x = tf.cast(i, tf.float32) x = tf.keras.applications.mobilenet.preprocess_input(x) core = tf.keras.applications.MobileNet() x = core(x) model = tf.keras.Model(inputs=[i], outputs=[x]) image = tf.image.decode_png(tf.io.read_file('file.png')) result = model(image) ``` Args: x: A floating point `numpy.array` or a `tf.Tensor`, 3D or 4D with 3 color channels, with values in the range [0, 255]. The preprocessed data are written over the input data if the data types are compatible. To avoid this behaviour, `numpy.copy(x)` can be used. data_format: Optional data format of the image tensor/array. Defaults to None, in which case the global setting `tf.keras.backend.image_data_format()` is used (unless you changed it, it defaults to "channels_last").{mode} Returns: Preprocessed `numpy.array` or a `tf.Tensor` with type `float32`. {ret} Raises: {error} a mode: One of "caffe", "tf" or "torch". Defaults to "caffe". - caffe: will convert the images from RGB to BGR, then will zero-center each color channel with respect to the ImageNet dataset, without scaling. - tf: will scale pixels between -1 and 1, sample-wise. - torch: will scale pixels between 0 and 1 and then will normalize each channel with respect to the ImageNet dataset. zE ValueError: In case of unknown `mode` or `data_format` argument.z; ValueError: In case of unknown `data_format` argument.zH The inputs pixel values are scaled between -1 and 1, sample-wise.z The input pixels values are scaled between 0 and 1 and each channel is normalized with respect to the ImageNet dataset.z The images are converted from RGB to BGR, then each color channel is zero-centered with respect to the ImageNet dataset, without scaling.z2keras.applications.imagenet_utils.preprocess_inputcaffec|dvrtd||tj}n|dvrtd|t|tjrt |||St|||S)z@Preprocesses a tensor or Numpy array encoding a batch of images.>tfrtorchzDExpected mode to be one of `caffe`, `tf` or `torch`. Received: mode=N> channels_lastchannels_firstz]Expected data_format to be one of `channels_first` or `channels_last`. Received: data_format=) data_formatmode) ValueErrorrimage_data_format isinstancenpndarray_preprocess_numpy_input_preprocess_symbolic_input)xr rs g/var/www/html/movieo_spanner_bot/venv/lib/python3.11/site-packages/keras/applications/imagenet_utils.pypreprocess_inputrgs +++ %" % %   /11 ? ? ? D6A D D   !RZ  Q&qkMMMM)!4PPPP)rreterrorz4keras.applications.imagenet_utils.decode_predictionsc0t|jdks|jddkr$tdt|jztXt jdtdd }t|5}tj |adddn #1swxYwYg}|D]c | dddd }fd |D}| d d | |d|S)aDecodes the prediction of an ImageNet model. Args: preds: Numpy array encoding a batch of predictions. top: Integer, how many top-guesses to return. Defaults to 5. Returns: A list of lists of top class prediction tuples `(class_name, class_description, score)`. One list of tuples per sample in batch input. Raises: ValueError: In case of invalid shape of the `pred` array (must be 2D). izx`decode_predictions` expects a batch of predictions (i.e. a 2D array of shape (samples, 1000)). Found array with shape: Nzimagenet_class_index.jsonmodels c2c37ea517e94d9795004a39431a14cb) cache_subdir file_hashcrg|]3}ttt||fz4S)tuple CLASS_INDEXstr).0ipreds r z&decode_predictions..s7OOOa% CFF+,,Qz9OOOrc|dS)Nrr')rs rz$decode_predictions..s !A$rT)keyreverse)lenshaperr*r)rget_fileCLASS_INDEX_PATHopenjsonloadargsortsortappend)predstopfpathfresults top_indicesresultr-s @rdecode_predictionsrDso& 5;1 A$ 6 6 '*-U[)9)9 :   # ' !8     %[[ 'A)A,,K ' ' ' ' ' ' ' ' ' ' ' ' ' ' 'GllnncTUU+DDbD1 OOOO;OOO  555v NsB##B'*B'ct|jjtjs(|t jd}|dkr |dz}|dz}|S|dkr|dz}gd}gd }nC|d kr*|jd kr|d d d df}n|d d d d d df}n |dd d d f}gd}d }|d kr|jd kr|dd d d d fxx|dzcc<|dd d d d fxx|dzcc<|dd d d d fxx|dzcc<|W|dd d d d fxx|dzcc<|dd d d d fxx|dzcc<|dd d d d fxx|dzcc<nI|d d dd d d d fxx|dzcc<|d d dd d d d fxx|dzcc<|d d dd d d d fxx|dzcc<|`|d d dd d d d fxx|dzcc<|d d dd d d d fxx|dzcc<|d d dd d d d fxx|dzcc<n|dxx|dzcc<|dxx|dzcc<|dxx|dzcc<|B|dxx|dzcc<|dxx|dzcc<|dxx|dzcc<|S)aPreprocesses a Numpy array encoding a batch of images. Args: x: Input array, 3D or 4D. data_format: Data format of the image array. mode: One of "caffe", "tf" or "torch". - caffe: will convert the images from RGB to BGR, then will zero-center each color channel with respect to the ImageNet dataset, without scaling. - tf: will scale pixels between -1 and 1, sample-wise. - torch: will scale pixels between 0 and 1 and then will normalize each channel with respect to the ImageNet dataset. Returns: Preprocessed Numpy array. F)copyr _@?r o@g ףp= ?gv/?gCl?gZd;O?gy&1?g?r Nr%.gjtY@g`"1]@gQ^@rr r).r).r ).r) issubclassdtypetyperfloatingastyperfloatxndim)rr rmeanstds rrrs( aglBK 0 03 HHW^%%EH 2 2 t|| U  S  U $$$### * * *v{{dddCiLaaa2slO#ttt) A)))&&& 6Q;; aAAAgJJJ$q' !JJJ aAAAgJJJ$q' !JJJ aAAAgJJJ$q' !JJJ!QQQ' c!f$ !QQQ' c!f$ !QQQ' c!f$ aaaAAAqqqjMMMT!W $MMM aaaAAAqqqjMMMT!W $MMM aaaAAAqqqjMMMT!W $MMM!!!Q111* Q' !!!Q111* Q' !!!Q111* Q'  & T!W  & T!W  & T!W ? fIIIQ III fIIIQ III fIIIQ III Hrc |dkr |dz}|dz}|S|dkr|dz}gd}gd}nP|dkr7tj|d kr|d d d d f}n|d d d d d d f}n |d d d d f}gd }d }tjtj| }tj|tj|kr=tj|tj|tj||}ntj|||}|Ztjtj|tj|}|dkrtj|d}||z}|S)aPreprocesses a tensor encoding a batch of images. Args: x: Input tensor, 3D or 4D. data_format: Data format of the image tensor. mode: One of "caffe", "tf" or "torch". - caffe: will convert the images from RGB to BGR, then will zero-center each color channel with respect to the ImageNet dataset, without scaling. - tf: will scale pixels between -1 and 1, sample-wise. - torch: will scale pixels between 0 and 1 and then will normalize each channel with respect to the ImageNet dataset. Returns: Preprocessed tensor. r rGrHr rIrJrKr rLNr%.rM)r )rO)r%r r ) rrTconstantrarrayrObias_addcastreshape)rr rrUrV mean_tensor std_tensors rrrs( t|| U  S  U $$$### * * *|A!##dddCiLaaa2slO#ttt) A)))"BHTNN?33K}Q7=5555   LgmA&6&6 7 7#     Q [ 9 9 %bhsmm7=;K;KLLL * * * Z@@J Z HrcV|dkr|rt|dkr|dkrE|ddvr/tjdt|dzdzd |d||f}nV|d dvr/tjdt|d zdzd |||d f}n|dkrd||f}n||df}|dkr!|r|||krt d |d ||S|r|dkr|t|dkrt d|ddkr|dkrt d|d|d |d|ks|d$|d|krt d|d|d|n|t|dkrt d|d dkr|dkrt d|d|d |d|ks|d$|d|krt d|d|d|n|r|}n |dkrd}nd}|rd |vrt d||S)aInternal utility to compute/validate a model's input shape. Args: input_shape: Either None (will return the default network input shape), or a user-provided shape to be validated. default_size: Default input width/height for the model. min_size: Minimum input width/height accepted by the model. data_format: Image data format to use. require_flatten: Whether the model is expected to be linked to a classifier via a Flatten layer. weights: One of `None` (random initialization) or 'imagenet' (pre-training on ImageNet). If weights='imagenet' input channels must be equal to 3. Returns: An integer shape tuple (may include None entries). Raises: ValueError: In case of invalid argument values. imagenetrLr r>r rLz]This model usually expects 1 or 3 input channels. However, it was passed an input_shape with z input channels.r) stacklevelr%NzXWhen setting `include_top=True` and loading `imagenet` weights, `input_shape` should be z. Received: input_shape=z0`input_shape` must be a tuple of three integers.z6The input must have 3 channels; Received `input_shape=`r zInput size must be at least rz; Received: input_shape=)rLNN)NNrLz[If `include_top` is True, you should specify a static `input_shape`. Received: input_shape=)r3warningswarnr*r) input_shape default_sizemin_sizer require_flattenweights default_shapes robtain_input_shaperk6s8*[1A1AQ1F1F * * *1~V++ B+a.))*)) ! )^\<HMM2f,, B+b/**+)) ! *<RIMM * * * l;MM)<;M*  "m++ ;/<;;.9;; .. * * *&{##q(($Jq>Q&&7j+@+@$7(3777  N.;q>H3L3L!!n0[^h5N5N$5x55$55'255 &{##q(($Jr?a''Gz,A,A$7(3777  N.;q>H3L3L!!n0[^h5N5N$5#55&.55'255  .'KK...- -  ;  7)477  rctjdkrdnd}tj|||dz}t|tr||f}|dd}nd|ddzz d|ddzz f}|ddz|ddzf}|d|dz |df|d|dz |dffS)zReturns a tuple for zero-padding for 2D convolution with downsampling. Args: inputs: Input tensor. kernel_size: An integer or tuple/list of 2 integers. Returns: A tuple. r rr rN)r r )rr int_shaperint)inputs kernel_sizeimg_dim input_sizeadjustcorrects r correct_padrus,..2BBBaaG"6**7gk+BCJ+s##1"K0 !}jma''Z]Q->)>?1~"KNa$78G fQi , fQi , rc|dStj|}|tjdtjdhvrtd|dS)a@validates that the classifer_activation is compatible with the weights. Args: classifier_activation: str or callable activation function weights: The pretrained weights to load. Raises: ValueError: if an activation other than `None` or `softmax` are used with pretrained weights. NsoftmaxzOnly `None` and `softmax` activations are allowed for the `classifier_activation` argument when using pretrained weights, with `include_top=True`; Received: classifier_activation=)rgetr)classifier_activationris rvalidate_activationrzs}'O,ABB ""% =&; = =   r)Nr)r)N)__doc__r8rcnumpyrkerasrr keras.utilsr tensorflow.python.util.tf_exportrr)r6PREPROCESS_INPUT_DOCPREPROCESS_INPUT_MODE_DOC"PREPROCESS_INPUT_DEFAULT_ERROR_DOCPREPROCESS_INPUT_ERROR_DOCPREPROCESS_INPUT_RET_DOC_TFPREPROCESS_INPUT_RET_DOC_TORCHPREPROCESS_INPUT_RET_DOC_CAFFErformatrDrrrkrurzr'rrrsGF """""":99999 % "H &H">K":"N BCCQQQDCQ,066 " ,7DEE(((FE(VF F F R9 9 9 D  vvvvr2     r