σ ΎχXc@@s²dZddlmZddlmZddlZddlmZddlmZddlmZdd lm Z dd lm Z dd lm Z dd lm Z dd lm Z ddlmZddlmZddlmZddlmZddlmZddlmZddlmZddlmZddlmZddlmZddlmZddlmZdZ dZ!d„Z"d"d„Z#e$ddddd d!„Z&dS(#s¨ResNet50 model for Keras. # Reference: - [Deep Residual Learning for Image Recognition](https://arxiv.org/abs/1512.03385) Adapted from code contributed by BigMoyan. i(tprint_function(tabsolute_importNi(tInput(tlayers(tDense(t Activation(tFlatten(tConv2D(t MaxPooling2D(t ZeroPadding2D(tAveragePooling2D(tGlobalAveragePooling2D(tGlobalMaxPooling2D(tBatchNormalization(tModel(tbackend(tget_source_inputs(t layer_utils(tget_filei(tdecode_predictions(tpreprocess_input(t_obtain_input_shapesvhttps://github.com/fchollet/deep-learning-models/releases/download/v0.2/resnet50_weights_tf_dim_ordering_tf_kernels.h5s|https://github.com/fchollet/deep-learning-models/releases/download/v0.2/resnet50_weights_tf_dim_ordering_tf_kernels_notop.h5c C@so|\}}}tjƒdkr*d}nd}dt|ƒ|d} dt|ƒ|d} t|dd| dƒ|ƒ} td |d| dƒ| ƒ} td ƒ| ƒ} t||d d d| d ƒ| ƒ} td |d| d ƒ| ƒ} td ƒ| ƒ} t|dd| dƒ| ƒ} td |d| dƒ| ƒ} tj| |gƒ} td ƒ| ƒ} | S(sμThe identity block is the block that has no conv layer at shortcut. # Arguments input_tensor: input tensor kernel_size: defualt 3, the kernel size of middle conv layer at main path filters: list of integers, the filterss of 3 conv layer at main path stage: integer, current stage label, used for generating layer names block: 'a','b'..., current block label, used for generating layer names # Returns Output tensor for the block. t channels_lastiitrest_branchtbntnamet2ataxistrelutpaddingtsamet2bt2c(ii(ii(tKtimage_data_formattstrRR RRtadd( t input_tensort kernel_sizetfilterststagetblocktfilters1tfilters2tfilters3tbn_axistconv_name_baset bn_name_basetx((s:/tmp/pip-build-isqEY4/keras/keras/applications/resnet50.pytidentity_block)s$   cC@sΉ|\}}}tjƒdkr*d} nd} dt|ƒ|d} dt|ƒ|d} t|dd|d| d ƒ|ƒ} td | d| d ƒ| ƒ} td ƒ| ƒ} t||d d d| dƒ| ƒ} td | d| dƒ| ƒ} td ƒ| ƒ} t|dd| dƒ| ƒ} td | d| dƒ| ƒ} t|dd|d| dƒ|ƒ} td | d| dƒ| ƒ} tj| | gƒ} td ƒ| ƒ} | S(snconv_block is the block that has a conv layer at shortcut # Arguments input_tensor: input tensor kernel_size: defualt 3, the kernel size of middle conv layer at main path filters: list of integers, the filterss of 3 conv layer at main path stage: integer, current stage label, used for generating layer names block: 'a','b'..., current block label, used for generating layer names # Returns Output tensor for the block. Note that from stage 3, the first conv layer at main path is with strides=(2,2) And the shortcut should have strides=(2,2) as well RiiRRRtstridesRRRRRRR R!t1(ii(ii(ii(R"R#R$RR RRR%(R&R'R(R)R*R3R+R,R-R.R/R0R1tshortcut((s:/tmp/pip-build-isqEY4/keras/keras/applications/resnet50.pyt conv_blockOs, timagenetiθc C@s—|dd:hkr!tdƒ‚n|dkrN|rN|dkrNtdƒ‚nt|ddddd tjƒd |ƒ}|d:kr–td |ƒ}n-tj|ƒs½td |d |ƒ}n|}tjƒd krήd}nd}td;ƒ|ƒ}tdd<dd=ddƒ|ƒ}t d|ddƒ|ƒ}t dƒ|ƒ}t d>dd?ƒ|ƒ}t |ddddgdddddd@ƒ}t |ddddgddddƒ}t |ddddgddddƒ}t |dddd gddddƒ}t |dddd gddddƒ}t |dddd gddddƒ}t |dddd gdddd!ƒ}t |dddd"gdd#ddƒ}t |dddd"gdd#ddƒ}t |dddd"gdd#ddƒ}t |dddd"gdd#dd!ƒ}t |dddd"gdd#dd$ƒ}t |dddd"gdd#dd%ƒ}t |dd d d&gdd'ddƒ}t |dd d d&gdd'ddƒ}t |dd d d&gdd'ddƒ}tdAdd(ƒ|ƒ}|r tƒ|ƒ}t|d)d*dd+ƒ|ƒ}n<|d,kr>tƒ|ƒ}n|d-kr\tƒ|ƒ}n|d:k rwt|ƒ} n|} t| |dd.ƒ} |dkr“|rΒtd/td0d1d2d3ƒ} ntd4td0d1d2d5ƒ} | j| ƒtjƒd6kr tj| ƒntjƒd7kr“|rk| jdd(ƒ} | jd} | jdd+ƒ}tj|| d7ƒntjƒd8krtj d9ƒqq“n| S(BsjInstantiates the ResNet50 architecture. Optionally loads weights pre-trained on ImageNet. Note that when using TensorFlow, for best performance you should set `image_data_format="channels_last"` in your Keras config at ~/.keras/keras.json. The model and the weights are compatible with both TensorFlow and Theano. The data format convention used by the model is the one specified in your Keras config file. # Arguments include_top: whether to include the fully-connected layer at the top of the network. weights: one of `None` (random initialization) or "imagenet" (pre-training on 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 (otherwise the input shape has to be `(224, 224, 3)` (with `channels_last` data format) or `(3, 224, 244)` (with `channels_first` data format). It should have exactly 3 inputs channels, and width and height should be no smaller than 197. E.g. `(200, 200, 3)` would be one valid value. 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. # Returns A Keras model instance. # Raises ValueError: in case of invalid argument for `weights`, or invalid input shape. R7spThe `weights` argument should be either `None` (random initialization) or `imagenet` (pre-training on ImageNet).iθsSIf using `weights` as imagenet with `include_top` as true, `classes` should be 1000t default_sizeiΰtmin_sizeiΕt data_formatt include_toptshapettensorRiii@iR3iRtconv1Rtbn_conv1RiR)R*tatbtci€itdiitetfiitavg_poolt activationtsoftmaxtfc1000tavgtmaxtresnet50s.resnet50_weights_tf_dim_ordering_tf_kernels.h5t cache_subdirtmodelstmd5_hasht a7b3fe01876f51b976af0dea6bc144ebs4resnet50_weights_tf_dim_ordering_tf_kernels_notop.h5t a268eb855778b3df3c7506639542a6afttheanotchannels_firstt tensorflowsςYou are using the TensorFlow backend, yet you are using the Theano image data format convention (`image_data_format="channels_first"`). For best performance, set `image_data_format="channels_last"` in your Keras config at ~/.keras/keras.json.N(ii(ii(ii(ii(ii(ii(ii(!tNonet ValueErrorRR"R#Rtis_keras_tensorR RR RRR6R2R RRR R RRRt WEIGHTS_PATHtWEIGHTS_PATH_NO_TOPt load_weightsRRtconvert_all_kernels_in_modelt get_layert output_shapet!convert_dense_weights_data_formattwarningstwarn(R;tweightsR&t input_shapetpoolingtclassest img_inputR.R1tinputstmodelt weights_pathtmaxpoolR<tdense((s:/tmp/pip-build-isqEY4/keras/keras/applications/resnet50.pytResNet50}s5     !-'''''''''''''''!        (ii('t__doc__t __future__RRR_RRtRRRRRR R R R R RNRRR"tengine.topologyRtutilsRtutils.data_utilsRtimagenet_utilsRRRRXRYR2R6tTrueRURk(((s:/tmp/pip-build-isqEY4/keras/keras/applications/resnet50.pyt s>  & .