2VhddlZddlZddlmZddlmZddlmZddlmZddlm Z ddl m Z ddl m Z dd lmZd Zd d d dddddZddddddddddddddddddddddddddddddddddddddddd d!dddddd!dd"d#dddgddddddddddd$ddddddddd$dd%ddddddd&d%d!ddddddd'd!d(ddddddd)d(d*ddddddd$d*d+dddddgddd,d,ddddddd&d,dddddddd&dd-ddddddd.d-d/ddddddd0d/d1ddddddd2d1d3ddddddd&d3d4dddddgddd,d5dddddddd5d,dddddddd,dddddddddd-ddddddd$d-d6ddddddd7d6d/dddddgddd,d5dddddddd5d,dddddddd,dddddddddd-ddddddd$d-d6ddddddd7d6d/dddddgddd,d5dddddddd5d,dddddddd,dddddddddd-ddddddd$d-d6ddddddd7d6d/dddddgddd,d5dddddddd5d,dddddddd,dddddddddd-ddddddd$d-d6ddddddd7d6d/dddddgd8Zd9d:d;dZd9d?d;d@d=d>ZdAZdBZdCZ dhdDZ dhdEZ didFZedGdHg djdIZedJdKg dkdLZedMdNg dldOZedPdQg dmdRZedSdTg dndUZedVdWg dodXZ edYdZg dpd[Z!ejEd\]e_#ejEd^]e_#ejEd_]e_#ejEd`]e_#ejEda]e_#ejEdb]e _#ejEdc]e!_#edddqdeZ$edfdrdgZ%e jJjFe%_#y)sN)backend) initializers)layers) keras_export)imagenet_utils) Functional)operation_utils) file_utilszMhttps://storage.googleapis.com/tensorflow/keras-applications/efficientnet_v2/) 21ecbf6da12460d5c40bb2f29ceb2188 893217f2bb855e2983157299931e43ff) 069f0534ff22adf035c89e2d9547a9dc 0e80663031ca32d657f9caa404b6ec37) 424e49f28180edbde1e94797771950a7 1dfe2e7a5d45b6632553a8961ea609eb) 1f1fc43bd98a6e4fd8fdfd551e02c7a0 f6abf7b5849ac99a89b50dd3fd532856) e1d88a8495beba45748fedd0cecbe016 af0682fb74e8c54910f2d4393339c070) a3bf6aa3276309f4fc6a34aa114c95cd 1b8dc055df72dde80d614482840fe342) 27e6d408b53c7ebc868fefa357689935 b0b66b5c863aef5b46e8608fe1711615)b0b1b2b3-s-m-l) kernel_size num_repeat input_filtersoutput_filters expand_ratiose_ratiostrides conv_type0@)r,r)r'r%r&r(r*r+g? Pi0i ` ip)efficientnetv2-sefficientnetv2-mefficientnetv2-lefficientnetv2-b0efficientnetv2-b1efficientnetv2-b2efficientnetv2-b3VarianceScalingg@fan_outtruncated_normal)scalemode distribution) class_nameconfiggUUUUUU?uniforma Instantiates the {name} architecture. Reference: - [EfficientNetV2: Smaller Models and Faster Training]( https://arxiv.org/abs/2104.00298) (ICML 2021) 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 EfficientNetV2, by default input preprocessing is included as a part of the model (as a `Rescaling` layer), and thus `keras.applications.efficientnet_v2.preprocess_input` is actually a pass-through function. In this use case, EfficientNetV2 models expect their inputs to be float tensors of pixels with values in the `[0, 255]` range. At the same time, preprocessing as a part of the model (i.e. `Rescaling` layer) can be disabled by setting `include_preprocessing` argument to `False`. With preprocessing disabled EfficientNetV2 models expect their inputs to be float tensors of pixels with values in the `[-1, 1]` range. Args: 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_shape: Optional shape tuple, only to be specified if `include_top` is `False`. It should have exactly 3 inputs channels. pooling: Optional pooling mode for feature extraction when `include_top` is `False`. Defaults to None. - `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. Defaults to 1000 (number of ImageNet classes). classifier_activation: A string 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. Defaults to `"softmax"`. When loading pretrained weights, `classifier_activation` can only be `None` or `"softmax"`. name: The name of the model (string). Returns: A model instance. cp||z}|xs|}t|t||dz z|z|z}t|S)z2Round number of filters based on depth multiplier.r!)maxint)filterswidth_coefficient min_depth depth_divisor minimum_depth new_filterss V/home/dcms/DCMS/lib/python3.12/site-packages/keras/src/applications/efficientnet_v2.py round_filtersraMsM   G.M Gma' '(M9MIK { cDttj||zS)z2Round number of repeats based on depth multiplier.)rYmathceilrepeatsdepth_coefficients r` round_repeatsriXs tyy*W45 66rbc  tjdk(rdnd tjd   f d} | S)z2MBConv block: Mobile Inverted Residual Bottleneck. channels_lastr r#block0c " z} dk7r~tj|ddtdtjd dz|}tj  dz|}tj  dz |}n|}tj tdtjd d z |}tj  d z|}tj  d z |}dcxkrdkrnntdt z}tj dz |}dk(r|ddf}ndd|f}tj| dz |}tj|ddt dz|}tj|dddt dz|}tj||g dz }tj ddtdtjd dz|}tj  dz|}dk(rD k(r?r!tjd dz|}tj||g dz }|S)Nr#sameF expand_convrZr%r+kernel_initializerpadding data_formatuse_biasname expand_bnaxismomentumruexpand_activationrudwconv2)r%r+depthwise_initializerrrrsrtrubn activationr se_squeeze se_reshape se_reducerrrrqrusigmoid se_expand se_excite project_conv project_bnNr#r#r#drop noise_shaperuadd)rConv2DCONV_KERNEL_INITIALIZERrimage_data_formatBatchNormalization ActivationDepthwiseConv2DrXrYGlobalAveragePooling2DReshapemultiplyDropoutrinputsrZx filters_sesese_shaperbn_axis bn_momentumr)r'r%rur(r*r+survival_probabilitys r`applyzMBConvBlock..applyos,. 1   #:#557M)  A))$K' A O!!*4:M3MNAA F " "#"9113 !    F % %;TD[   DF  jtl/B CA F x 1 QMH$< =>JH..D<4GHKB!|#Q?q'?Ctl/BCBGB%#:K' B$#:K' BBd[.@AA FMM"6113&    F % %;TL5H   a.applys,. 1   '#:#557M)  A)){ 9KA!!%D3F,FAA x 1 QMH$< =>JH..D<4GHKB!|#Q?q'?Ctl/BCBGB%#:K' B$#:K' BBd[.@AA FMM )Q.K%*A6&    F % %;TL5H   1 !!%D3G,GA ad<*|}|dkDrtj@|d=*|}t jB|| tjD||tFtIjJd>d?@|}nC|dAk(rtj>d<*|}n!|dBk(rtjLdC*|}| tOjP| }#n|}#tS|#|| *}$| dk(rW| rdD}%tT|d}&ndE}%tT|d }&| |%z}'tjV|'tX|'zdF|&G}(|$j[|(|$S| |$j[| |$S)Ia Instantiates the EfficientNetV2 architecture using given scaling coefficients. Args: width_coefficient: float, scaling coefficient for network width. depth_coefficient: float, scaling coefficient for network depth. default_size: integer, default input image size. dropout_rate: float, dropout rate before final classifier layer. drop_connect_rate: float, dropout rate at skip connections. depth_divisor: integer, a unit of network width. min_depth: integer, minimum number of filters. bn_momentum: float. Momentum parameter for Batch Normalization layers. activation: activation function. blocks_args: list of dicts, parameters to construct block modules. name: string, model name. include_top: whether to include the fully-connected layer at the top of the network. weights: one of `None` (random initialization), `"imagenet"` (pre-training on ImageNet), or the path to the weights file to be loaded. input_tensor: optional Keras tensor (i.e. output of `layers.Input()`) or numpy array to use as image input for the model. input_shape: optional shape tuple, only to be specified if `include_top` is `False`. It should have exactly 3 inputs channels. 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 string 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. include_preprocessing: Boolean, whether to include the preprocessing layer (`Rescaling`) at the bottom of the network. Defaults to `True`. Returns: A model instance. default>NimagenetzThe `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=rzRIf using `weights="imagenet"` with `include_top` as true, `classes` should be 1000r<) default_sizemin_sizersrequire_flattenweights)shape)tensorrrkr r#-bgp?)rQchannels_first ףp= ?v/?Cl?g٪?.?Q?)rrr)rrr)meanvariancerxg?)rQoffsetrr')rZr[r\r]r!rnF stem_conv)rZr%r+rqrrrtrustem_bnrwstem_activationr{c3&K|] }|d yw)r&N).0argss r` z!EfficientNetV2..sBdtL)Bsr&r()rr#r,rfr+blocka_)rrrruitop_convrptop_bntop_activationravg_pool top_dropoutr$ predictions)rrqbias_initializerruavgrXmax_poolz.h5z _notop.h5models) cache_subdir file_hashr).DEFAULT_BLOCKS_ARGSr exists ValueErrorrobtain_input_shaperrrInputis_keras_tensorsplit startswith Rescaling Normalizationrarrrrcopydeepcopyfloatsum enumeraterrpoprirangechrrrvalidate_activationDenseDENSE_KERNEL_INITIALIZERrConstantGlobalMaxPooling2Dr get_source_inputsrWEIGHTS_HASHESget_fileBASE_WEIGHTS_PATH load_weights))r[rhr dropout_ratedrop_connect_rater]r\rr blocks_argsru include_topr input_tensor input_shapepoolingclassesclassifier_activationinclude_preprocessing weights_name img_inputrr num_channelsrr stem_filtersrblocksirrrgj top_filtersrmodel file_suffixr file_name weights_paths) r`EfficientNetV2r 7sJi)$/ ) )Z->->w-G !")  +  *D 1  !33!--/# KLL{3 &&|4 L LI$I,,./AaqGA#7Q;/ ::c?2  ) )# .<13D1  y1!4A((*.>> 'eWIy9:j\XJ<8*>-9$$! A ?  {2>qAA!A/+# L   2   A !!     A >*+<=a@A-- ,K A 3BkBB CF[)$4L!A%%%!.)/' ! _ "/)*/' "   $45dhh{6KLHH\*>O w A1u"#Y(,-=(>_%%'%6%:V%CQUGCBK=2     A FA -$N +# K  2--/   A !!     A HZ6FGJA :F ) )z :1 = ! @|-@CA**+@'J FLL ,7)2237    e >--:>qAA  :))z:1=A 22<@ vqt ,E* K&|4Q7I%K&|4Q7I;& !**   )!  <( L   7# Lrbz3keras.applications.efficientnet_v2.EfficientNetV2B0z#keras.applications.EfficientNetV2B0c 2tddd|||||||||d S)N?rAr r[rhrrurrrrrrrrrr  rrrrrrrrrus r`EfficientNetV2B0rF8"  !33 rbz3keras.applications.efficientnet_v2.EfficientNetV2B1z#keras.applications.EfficientNetV2B1c 2tddd|||||||||d S)Nr 皙?rrrrs r`EfficientNetV2B1rhrrbz3keras.applications.efficientnet_v2.EfficientNetV2B2z#keras.applications.EfficientNetV2B2c 2tddd|||||||||d S)Nr333333?irrrrs r`EfficientNetV2B2rrrbz3keras.applications.efficientnet_v2.EfficientNetV2B3z#keras.applications.EfficientNetV2B3c 2tddd|||||||||d S)Nrgffffff?i,rrrrs r`EfficientNetV2B3rrrbz2keras.applications.efficientnet_v2.EfficientNetV2Sz"keras.applications.EfficientNetV2Sc 2tddd|||||||||d S)Nr rCrrrrs r`EfficientNetV2Srrrbz2keras.applications.efficientnet_v2.EfficientNetV2Mz"keras.applications.EfficientNetV2Mc 2tddd|||||||||d S)Nr rrrrs r`EfficientNetV2Mr rrbz2keras.applications.efficientnet_v2.EfficientNetV2Lz"keras.applications.EfficientNetV2Lc 2tddd|||||||||d S)Nr rrrrrs r`EfficientNetV2Lr"rrbrr{rrrrr r"z3keras.applications.efficientnet_v2.preprocess_inputc|S)aA placeholder method for backward compatibility. The preprocessing logic has been included in the EfficientNetV2 model implementation. Users are no longer required to call this method to normalize the input data. This method does nothing and only kept as a placeholder to align the API surface between old and new version of model. Args: x: A floating point `numpy.array` or a tensor. data_format: Optional data format of the image tensor/array. 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").{mode} Returns: Unchanged `numpy.array` or tensor. r)rrss r`preprocess_inputr$=s & Hrbz5keras.applications.efficientnet_v2.decode_predictionsc0tj||S)N)top)rdecode_predictions)predsr&s r`r'r'Ss  , ,U <r7sg  "-1')&c       3D      !  !  ! g=@      !  !  !  ! {GT      !  !  !  ! {GT        !  ! g=@        !  ! g=@        !  ! g=@        !  ! g=S GT$*$!AH7  qn  cT    #)L^=-   #   8=-   #   8=-   #   8=-   #   8<,   #   8<,   #   8<,   #   8*006H0I)006H0I)006H0I)006H0I(//5F/G(//5F/G(//5F/GCD E *EF=G=,>>FFrb