&VfFddlZddlmZddlmZddlmZddlmZddlm Z ddl m Z ddl m Z d Zed d g ddZdZddZedddZedddZejdejeje_ejje_dS)N)backend)layers) keras_export)imagenet_utils) Functional)operation_utils) file_utilszJhttps://storage.googleapis.com/tensorflow/keras-applications/mobilenet_v2/z+keras.applications.mobilenet_v2.MobileNetV2zkeras.applications.MobileNetV2?Timagenetsoftmaxc |dvs'tj|std|d|dkr|r|dkrtd|d|| tj|}ni#t$r\ tjt j|}n1#t$r$td|d t|dwxYwYnwxYw|r}tjd kr3|j d |d krtd |j d |dnF|j d|d krtd|j d |dntd|d|| tj|n0#t$r#td|dt|wxYw|tj|sd} n|otj|r[tjd kr|j d} |j d} n|j d } |j d} | | kr| dvr| } nOd} nL|d} nGtjd kr|d } |d} n|d} |d } | | kr| dvr| } nd} tj || dtj||}tjdkrd\} } nd\} } || } || } |dkr9|dvrtd|d| | ks| dvrd} tj dd|tj| }n-tj|stj||!}n|}tjd krd nd"}t!d|zd#}tj|dd$d%d&d'(|}tj|d)d*d+,|}tjd-d./|}t)|d0|d d d1}t)|d2|dd3d 1}t)|d2|d d3d1}t)|d|dd3d1}t)|d|d d3d41}t)|d|d d3d51}t)|d6|dd3d31}t)|d6|d d3d71}t)|d6|d d3d#1}t)|d6|d d3d81}t)|d9|d d3d:1}t)|d9|d d3d;1}t)|d9|d d3d<1}t)|d=|dd3d>1}t)|d=|d d3d?1}t)|d=|d d3d@1}t)|dA|d d3d01}|dBkrt!dC|zd#}ndC}tj|d d&dDE|}tj|d)d*dF,|}tjd-dG/|}|rRtj|}tj||tj||dHI|}nE|dJkrtj|}n"|dKkrtj|}|t j|}n|}t3||dL|dMdN| /}|dkr|rWdOt5t7|zdNzt5| zdPz}t8|z}tj||dQR}nYdOt5t7|zdNzt5| zdSzdPz}t8|z}tj||dQR}||n||||S)TaInstantiates the MobileNetV2 architecture. MobileNetV2 is very similar to the original MobileNet, except that it uses inverted residual blocks with bottlenecking features. It has a drastically lower parameter count than the original MobileNet. MobileNets support any input size greater than 32 x 32, with larger image sizes offering better performance. Reference: - [MobileNetV2: Inverted Residuals and Linear Bottlenecks]( https://arxiv.org/abs/1801.04381) (CVPR 2018) This function returns a Keras image classification model, optionally loaded with weights pre-trained on ImageNet. For image classification use cases, see [this page for detailed examples]( https://keras.io/api/applications/#usage-examples-for-image-classification-models). For transfer learning use cases, make sure to read the [guide to transfer learning & fine-tuning]( https://keras.io/guides/transfer_learning/). Note: each Keras Application expects a specific kind of input preprocessing. For MobileNetV2, call `keras.applications.mobilenet_v2.preprocess_input` on your inputs before passing them to the model. `mobilenet_v2.preprocess_input` will scale input pixels between -1 and 1. Args: input_shape: Optional shape tuple, only to be specified if `include_top` is `False` (otherwise the input shape has to be `(224, 224, 3)` (with `"channels_last"` data format) or `(3, 224, 224)` (with `"channels_first"` data format). It should have exactly 3 inputs channels, and width and height should be no smaller than 32. E.g. `(200, 200, 3)` would be one valid value. Defaults to `None`. `input_shape` will be ignored if the `input_tensor` is provided. alpha: Controls the width of the network. This is known as the width multiplier in the MobileNet paper. - If `alpha < 1.0`, proportionally decreases the number of filters in each layer. - If `alpha > 1.0`, proportionally increases the number of filters in each layer. - If `alpha == 1`, default number of filters from the paper are used at each layer. Defaults to `1.0`. include_top: Boolean, whether to include the fully-connected layer at the top of the network. Defaults to `True`. weights: One of `None` (random initialization), `"imagenet"` (pre-training on ImageNet), or the path to the weights file to be loaded. Defaults to `"imagenet"`. input_tensor: Optional Keras tensor (i.e. output of `layers.Input()`) to use as image input for the model. `input_tensor` is useful for sharing inputs between multiple different networks. Defaults to `None`. pooling: Optional pooling mode for feature extraction when `include_top` is `False`. - `None` (default) means that the output of the model will be the 4D tensor output of the last convolutional block. - `avg` means that global average pooling will be applied to the output of the last convolutional block, and thus the output of the model will be a 2D tensor. - `max` means that global max pooling will be applied. classes: Optional number of classes to classify images into, only to be specified if `include_top` is `True`, and if no `weights` argument is specified. Defaults to `1000`. classifier_activation: A `str` or callable. The activation function to use on the "top" layer. Ignored unless `include_top=True`. Set `classifier_activation=None` to return the logits of the "top" layer. When loading pretrained weights, `classifier_activation` can only be `None` or `"softmax"`. Returns: A model instance. >Nr zThe `weights` argument should be either `None` (random initialization), `imagenet` (pre-training on ImageNet), or the path to the weights file to be loaded. Received `weights=`r r zfIf using `weights="imagenet"` with `include_top` as true, `classes` should be 1000. Received `classes=Nzinput_tensor: z7is not type input_tensor. Received `type(input_tensor)=channels_firstz|input_shape[1] must equal shape(input_tensor)[1] when `image_data_format` is `channels_first`; Received `input_tensor.shape=z`, `input_shape=zNinput_tensor.shape[2] must equal input_shape[1]; Received `input_tensor.shape=z;input_tensor is not a Keras tensor; Received `input_tensor=z9input_tensor must be a valid Keras tensor type; Received z of type )`rr ) default_sizemin_size data_formatrequire_flattenweights channels_last)rr)rr)gffffff?g?g?r g?gffffff?zIf imagenet weights are being loaded, alpha must be one of `0.35`, `0.50`, `0.75`, `1.0`, `1.3` or `1.4` only; Received `alpha=z`input_shape` is undefined or non-square, or `rows` is not in [96, 128, 160, 192, 224]. Weights for input shape (224, 224) will be loaded as the default.) stacklevel)shape)tensorr!)rrsameFConv1) kernel_sizestridespaddinguse_biasnameMbP?+?bn_Conv1axisepsilonmomentumr+@ Conv1_relur+)filtersalphastride expansionblock_id@ r r i@r iConv_1)r'r*r+ Conv_1_bnout_relu predictions) activationr+avgmax mobilenetv2_z0.2f_0mobilenet_v2_weights_tf_dim_ordering_tf_kernels_z.h5models) cache_subdir_no_top)r exists ValueErrorris_keras_tensorrget_source_inputstypeimage_data_formatr!robtain_input_shapewarningswarnrInput_make_divisibleConv2DBatchNormalizationReLU_inverted_res_blockGlobalAveragePooling2Dvalidate_activationDenseGlobalMaxPooling2DrstrfloatBASE_WEIGHT_PATHget_file load_weights) input_shaper8 include_topr input_tensorpoolingclassesclassifier_activationis_input_t_tensorrrowscolsrow_axiscol_axis img_input channel_axisfirst_block_filtersxlast_block_filtersinputsmodel model_name weight_path weights_paths `/var/www/html/software/conda/lib/python3.11/site-packages/keras/src/applications/mobilenet_v2.py MobileNetV2rs | ) ) )Z->w-G-G ) ,")  , , ,   *D ODK O O O   <#;  ' 7 E E      $+$;#5lCC%%!!    J\JJ484F4FJJJ "!   (**.>>>%a(KN::$9(-99+6 999; %a(KN::$7'-77)4777;:*6:::  |7   #L 1 1 1 1   H(HH37 3E3EHH    w'>|'L'L LL  W%<\%J%J (**.>>>#)!,#)!,#)!,#)!,t||(@ @ @# "      $ & &*: : :q>Dq>DDq>Dq>D 4<>-F+c tjdkrdnd}|j|}t||z}t |d} |} d|d} |rpt j||zdddd | d z | } t j|d d | dz| } t jd| dz| } nd} |dkr5t j tj | d| dz| } t j d|d d|dkrdnd| dz| } t j|d d | dz| } t jd| dz| } t j| dddd | dz | } t j|d d | dz| } || kr)|dkr#t j | dz|| gS| S) zInverted ResNet block.rrr#r$block_rQr%FNexpand)r'r)r*rMr+r,r- expand_BNr/r3 expand_relur5expanded_conv_rrpad)r)r+valid depthwise)r'r(rMr*r)r+ depthwise_BNdepthwise_reluproject project_BNadd)rr[r!intr`rrarbrc ZeroPadding2Dr correct_padDepthwiseConv2DAdd) r~r:r9r8r7r;rz in_channelspointwise_conv_filterspointwise_filtersr|prefixs rrdrds1337GGG11RL,|,K 511((>BBA !h ! ! !F" FM  #("        F %+%       :FK&="8 9 9 9! < <!{{ F ".q!446E>       A++7 k !      A !  n $        A 9 Cf'77888;;A   i       A !  l "        A'''FaKK.vzv~...{;;; Hc|||}t|t||dz z|z|z}|d|zkr||z }|S)Nrg?)rOr)vdivisor min_valuenew_vs rr`r`sR  3q7Q;//7:WD E EE sQw  Lrz0keras.applications.mobilenet_v2.preprocess_inputc0tj||dS)Ntf)rmode)rpreprocess_input)r|rs rrrs#  * {   rz2keras.applications.mobilenet_v2.decode_predictionsr?c.tj||S)N)top)rdecode_predictions)predsrs rrrs  ,U < < <rs------111111''''''))))))&&&&&&Q 5(    #uuu  up H H H V@AABA BCC===DC=*>EE 2  3F ,>Fr