o i e@s|dZddlmmZddlmZddlmZddl m Z ddl m Z ddl mZddl mZdd lmZdd lmZd Zed Zed ZedZedZe Z             d5ddZedd       d6ddZedd       d6d d!Z d7d$d%Zd8d&d'Zd8d(d)Z d8d*d+Z!ed,d8d-d.Z"ed/d9d1d2Z#ej$j%d3ej&ej'd4e"_ej#je#_dS):aGNASNet-A models for Keras. NASNet refers to Neural Architecture Search Network, a family of models that were designed automatically by learning the model architectures directly on the dataset of interest. Here we consider NASNet-A, the highest performance model that was found for the CIFAR-10 dataset, and then extended to ImageNet 2012 dataset, obtaining state of the art performance on CIFAR-10 and ImageNet 2012. Only the NASNet-A models, and their respective weights, which are suited for ImageNet 2012 are provided. The below table describes the performance on ImageNet 2012: --------------------------------------------------------------------------- Architecture | Top-1 Acc | Top-5 Acc | Multiply-Adds | Params (M) ---------------------|-----------|-----------|----------------|------------ NASNet-A (4 @ 1056) | 74.0 % | 91.6 % | 564 M | 5.3 NASNet-A (6 @ 4032) | 82.7 % | 96.2 % | 23.8 B | 88.9 Reference: - [Learning Transferable Architectures for Scalable Image Recognition]( https://arxiv.org/abs/1707.07012) (CVPR 2018) N)backend)imagenet_utils)training)VersionAwareLayers) data_utils) layer_utils) tf_logging) keras_exportzDhttps://storage.googleapis.com/tensorflow/keras-applications/nasnet/zNASNet-mobile.h5zNASNet-mobile-no-top.h5zNASNet-large.h5zNASNet-large-no-top.h5`Timagenetsoftmaxc  Cs|dvstjj|std|dkr|r| dkrtdt|tr4d|vr4|dkr4tdt|d| dur:d } tj || d t ||d }t d krZt d t d d} nd} |durgtj|d}nt |sttj||d}n|}|d|ddkrtd|t dkrdnd}|d}tj|ddddddd|}tj|ddd d!|}d}t||||dd"d#\}}t||||d$d#\}}t|D]}t|||d%|d#\}}qt||||d&|d#\}}|s|n|}t|D]}t||||d%||dd#\}}qt||||dd&d|d#\}}|s |n|}t|D]}t||||dd%d||dd#\}}q&td'|}|rat|}t| |tj| | d(d)|}n| d*krmt|}n | d+krxt|}|durt|}n|}tj||d,d-}|dkr| d.kr|rt j!d/t"d0d1d2}n t j!d3t#d0d4d2}|$|n/| d kr|rt j!d5t%d0d6d2}n t j!d7t&d0d8d2}|$|ntd9|dur|$|| rt | |S):aInstantiates a NASNet model. Reference: - [Learning Transferable Architectures for Scalable Image Recognition]( https://arxiv.org/abs/1707.07012) (CVPR 2018) 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 NasNet, call `tf.keras.applications.nasnet.preprocess_input` on your inputs before passing them to the model. `nasnet.preprocess_input` will scale input pixels between -1 and 1. Args: input_shape: Optional shape tuple, the input shape is by default `(331, 331, 3)` for NASNetLarge and `(224, 224, 3)` for NASNetMobile. It should have exactly 3 input channels, and width and height should be no smaller than 32. E.g. `(224, 224, 3)` would be one valid value. penultimate_filters: Number of filters in the penultimate layer. NASNet models use the notation `NASNet (N @ P)`, where: - N is the number of blocks - P is the number of penultimate filters num_blocks: Number of repeated blocks of the NASNet model. NASNet models use the notation `NASNet (N @ P)`, where: - N is the number of blocks - P is the number of penultimate filters stem_block_filters: Number of filters in the initial stem block skip_reduction: Whether to skip the reduction step at the tail end of the network. filter_multiplier: Controls the width of the network. - If `filter_multiplier` < 1.0, proportionally decreases the number of filters in each layer. - If `filter_multiplier` > 1.0, proportionally increases the number of filters in each layer. - If `filter_multiplier` = 1, default number of filters from the paper are used at each layer. include_top: Whether to include the fully-connected layer at the top of the network. weights: `None` (random initialization) or `imagenet` (ImageNet weights) input_tensor: Optional Keras tensor (i.e. output of `layers.Input()`) to use as image input for the model. pooling: Optional pooling mode for feature extraction when `include_top` is `False`. - `None` 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. default_size: Specifies the default image size of the model 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 `keras.Model` instance. >NrzThe `weights` argument should be either `None` (random initialization), `imagenet` (pre-training on ImageNet), or the path to the weights file to be loaded.rrzWIf using `weights` as `"imagenet"` with `include_top` as true, `classes` should be 1000NzWhen specifying the input shape of a NASNet and loading `ImageNet` weights, the input_shape argument must be static (no None entries). Got: `input_shape=z`.K ) default_sizemin_size data_formatrequire_flattenweights channels_lastaThe NASNet family of models is only available for the input data format "channels_last" (width, height, channels). However your settings specify the default data format "channels_first" (channels, width, height). You should set `image_data_format="channels_last"` in your Keras config located at ~/.keras/keras.json. The model being returned right now will expect inputs to follow the "channels_last" data format.channels_first)shape)tensorrr rzwFor NASNet-A models, the `penultimate_filters` must be a multiple of 24 * (`filter_multiplier` ** 2). Current value: %dr r r validF stem_conv1 he_normal)stridespaddinguse_biasnamekernel_initializerZڊ?MbP?stem_bn1axismomentumepsilonr(stem_1block_idstem_2z%dz reduce_%drelu predictions) activationr(avgmaxNASNetr(znasnet_mobile.h5models 020fb642bf7360b370c678b08e0adf61) cache_subdir file_hashznasnet_mobile_no_top.h5 1ed92395b5b598bdda52abe5c0dbfd63znasnet_large.h5 11577c9a518f0070763c2b964a382f17znasnet_large_no_top.h5 d81d89dc07e6e56530c4e77faddd61b5zDImageNet weights can only be loaded with NASNetLarge or NASNetMobile)'tfiogfileexists ValueError isinstancetuplestrrobtain_input_shaperimage_data_formatloggingwarningset_image_data_formatlayersInputis_keras_tensorConv2DBatchNormalization_reduction_a_cellrange_normal_a_cell ActivationGlobalAveragePooling2Dvalidate_activationDenseGlobalMaxPooling2Drget_source_inputsrModelrget_fileNASNET_MOBILE_WEIGHT_PATH NASNET_MOBILE_WEIGHT_PATH_NO_TOP load_weightsNASNET_LARGE_WEIGHT_PATHNASNET_LARGE_WEIGHT_PATH_NO_TOP) input_shapepenultimate_filters num_blocksstem_block_filtersskip_reductionfilter_multiplier include_topr input_tensorpoolingclassesrclassifier_activationold_data_format img_input channel_dimfiltersxpip0inputsmodel weights_pathr|L/var/www/myenv/lib/python3.10/site-packages/keras/src/applications/nasnet.pyr:@s6[                              r:z&keras.applications.nasnet.NASNetMobilezkeras.applications.NASNetMobilecC"t|ddddd|||||d|d S)a Instantiates a Mobile NASNet model in ImageNet mode. Reference: - [Learning Transferable Architectures for Scalable Image Recognition]( https://arxiv.org/abs/1707.07012) (CVPR 2018) Optionally loads weights pre-trained on ImageNet. Note that the data format convention used by the model is the one specified in your Keras config at `~/.keras/keras.json`. Note: each Keras Application expects a specific kind of input preprocessing. For NASNet, call `tf.keras.applications.nasnet.preprocess_input` on your inputs before passing them to the model. 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)` for NASNetMobile It should have exactly 3 inputs channels, and width and height should be no smaller than 32. E.g. `(224, 224, 3)` would be one valid value. include_top: Whether to include the fully-connected layer at the top of the network. weights: `None` (random initialization) or `imagenet` (ImageNet weights). For loading `imagenet` weights, `input_shape` should be (224, 224, 3) input_tensor: Optional Keras tensor (i.e. output of `layers.Input()`) to use as image input for the model. pooling: Optional pooling mode for feature extraction when `include_top` is `False`. - `None` means that the output of the model will be the 4D tensor output of the last convolutional layer. - `avg` means that global average pooling will be applied to the output of the last convolutional layer, 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. 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 Keras model instance. Raises: ValueError: In case of invalid argument for `weights`, or invalid input shape. RuntimeError: If attempting to run this model with a backend that does not support separable convolutions. i rFr r< rgrhrirjrkrlrrmrnrorrpr:rfrlrrmrnrorpr|r|r} NASNetMobiledGrz%keras.applications.nasnet.NASNetLargezkeras.applications.NASNetLargecCr~)a Instantiates a NASNet model in ImageNet mode. Reference: - [Learning Transferable Architectures for Scalable Image Recognition]( https://arxiv.org/abs/1707.07012) (CVPR 2018) Optionally loads weights pre-trained on ImageNet. Note that the data format convention used by the model is the one specified in your Keras config at `~/.keras/keras.json`. Note: each Keras Application expects a specific kind of input preprocessing. For NASNet, call `tf.keras.applications.nasnet.preprocess_input` on your inputs before passing them to the model. Args: input_shape: Optional shape tuple, only to be specified if `include_top` is False (otherwise the input shape has to be `(331, 331, 3)` for NASNetLarge. It should have exactly 3 inputs channels, and width and height should be no smaller than 32. E.g. `(224, 224, 3)` would be one valid value. include_top: Whether to include the fully-connected layer at the top of the network. weights: `None` (random initialization) or `imagenet` (ImageNet weights). For loading `imagenet` weights, `input_shape` should be (331, 331, 3) input_tensor: Optional Keras tensor (i.e. output of `layers.Input()`) to use as image input for the model. pooling: Optional pooling mode for feature extraction when `include_top` is `False`. - `None` means that the output of the model will be the 4D tensor output of the last convolutional layer. - `avg` means that global average pooling will be applied to the output of the last convolutional layer, 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. 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 Keras model instance. Raises: ValueError: in case of invalid argument for `weights`, or invalid input shape. RuntimeError: If attempting to run this model with a backend that does not support separable convolutions. r r r Tr rrrrr|r|r} NASNetLargerrrrrc Cstdkrdnd}td|mtd|}|dkr1tjt||d|d|}d }nd }tj|||d ||d d d|}tj |ddd|d|}td|}tj||d|d d d d|}tj |ddd|d|}Wd|S1swY|S)aFAdds 2 blocks of [relu-separable conv-batchnorm]. Args: ip: Input tensor filters: Number of output filters per layer kernel_size: Kernel size of separable convolutions strides: Strided convolution for downsampling block_id: String block_id Returns: A Keras tensor rrrseparable_conv_block_r5r!separable_conv_1_pad_r&r(r"sameseparable_conv_1_Fr$)r%r(r&r'r)r*r+separable_conv_1_bn_r-separable_conv_2_)r(r&r'r)separable_conv_2_bn_N) rrM name_scoperQrY ZeroPadding2Dr correct_padSeparableConv2DrU)iprt kernel_sizer%r3rsruconv_padr|r|r}_separable_conv_blocksl   ((rc Csztdkrdnd}tdkrdnd}t|}|dur"t|}td |dur0|}n||||krtd|{tjd d |d |}tjd d dd|d|}tj|dd ddd|dd|}tjdd|} tj dd| } tjd d dd|d| } tj|dd ddd|dd| } tj || g|d}tj |ddd|d |}Wdn1swYn^|||krtd!|.td |}tj|d d dd"|ddd#|}tj |ddd|d |}Wdn)1swYWd|SWd|SWd|SWd|SWd|S1s6wY|S)$aAdjusts the input `previous path` to match the shape of the `input`. Used in situations where the output number of filters needs to be changed. Args: p: Input tensor which needs to be modified ip: Input tensor whose shape needs to be matched filters: Number of output filters to be matched block_id: String block_id Returns: Adjusted Keras tensor rrrr N adjust_blockadjust_reduction_block_r5adjust_relu_1_r;rr!r"adjust_avg_pool_1_r%r&r(rFadjust_conv_1_r$)r&r'r(r))rrr)r&)rrr)croppingadjust_avg_pool_2_adjust_conv_2_)r.r*r+ adjust_bn_r-adjust_projection_block_adjust_conv_projection_r%r&r(r'r)) rrM int_shaperrQrYAveragePooling2DrTr Cropping2D concatenaterU) rvrrtr3rsimg_dimip_shapep_shapep1p2r|r|r} _adjust_blockPs    ,  C C C C CCrc Cstdkrdnd}td|Ot||||}td|}tj|dddd|d d d |}tj|d d d|d|}td)t||dd|d}t||d|d}tj ||gd|d}Wdn1srwYtd*t||dd|d} t||dd|d} tj | | gd|d} Wdn1swYtd"tj dddd|d |} tj | |gd!|d} Wdn1swYtd"0tj dddd#|d |} tj dddd$|d |}tj | |gd%|d}Wdn 1swYtd&t||d'|d}tj ||gd(|d}Wdn 1sBwYtj ||| | ||g|d)|d*}Wd||fS1sfwY||fS)+zAdds a Normal cell for NASNet-A (Fig. 4 in the paper). Args: ip: Input tensor `x` p: Input tensor `p` filters: Number of output filters block_id: String block_id Returns: A Keras tensor rrrnormal_A_block_r5rrnormal_conv_1_Fr$rr*r+ normal_bn_1_r-block_1r normal_left1_)rr3normal_right1_r2 normal_add_1_r;Nblock_2 normal_left2_rnormal_right2_ normal_add_2_block_3 normal_left3_r normal_add_3_block_4 normal_left4_normal_right4_ normal_add_4_block_5 normal_left5_ normal_add_5_normal_concat_r.r() rrMrrrQrYrTrUraddrr)rrvrtr3rshx1_1x1_2x1x2_1x2_2x2x3x4_1x4_2x4x5rur|r|r}rXs          LLrXc Cstdkrdnd}td|mt||||}td|}tj|dddd|d d d |}tj|d d d|d|}tjt |dd|d|}td,t ||ddd|d}t ||ddd|d}tj ||gd|d} Wdn1swYtd.tj dddd |d!|} t ||ddd"|d} tj | | gd#|d} Wdn1swYtd$.tjdddd%|d!|} t ||ddd&|d}tj | |gd'|d}Wdn1swYtd(tjdddd)|d!| }t | |g}Wdn 1s&wYtd*-t | |dd)|d+}tj dddd,|d!|}tj ||gd-|d}Wdn 1sawYtj| |||g|d.|d/}||fWdS1swYdS)0zAdds a Reduction cell for NASNet-A (Fig. 4 in the paper). Args: ip: Input tensor `x` p: Input tensor `p` filters: Number of output filters block_id: String block_id Returns: A Keras tensor rrrreduction_A_block_r5rrreduction_conv_1_Fr$rr*r+reduction_bn_1_r-r reduction_pad_1_rrrr!reduction_left1_)r%r3)rreduction_right1_reduction_add_1_r;Nrrr"reduction_left2_rreduction_right2_reduction_add_2_rreduction_left3_reduction_right3_reduction_add3_rreduction_left4_rr2reduction_right5_reduction_add4_reduction_concat_r)rrMrrrQrYrTrUrrrrr MaxPooling2Drr)rrvrtr3rsrh3rrrrrrx3_1x3_2rrx5_1x5_2rrur|r|r}rV s         &rVz*keras.applications.nasnet.preprocess_inputcCstj||ddS)NrD)rmode)rpreprocess_input)rurr|r|r}r}srz,keras.applications.nasnet.decode_predictionsrcCstj||dS)N)top)rdecode_predictions)predsrr|r|r}rsr)rreterror) Nr r r Tr TrNNrNr)NTrNNrr)rrN)N)r)(__doc__tensorflow.compat.v2compatv2rD keras.srcrkeras.src.applicationsrkeras.src.enginerkeras.src.layersrkeras.src.utilsrrtensorflow.python.platformrrN tensorflow.python.util.tf_exportr BASE_WEIGHTS_PATHrarbrdrerQr:rrrrrXrVrrPREPROCESS_INPUT_DOCformatPREPROCESS_INPUT_RET_DOC_TFPREPROCESS_INPUT_ERROR_DOCr|r|r|r}s         & U V < \ ]t