Python keras.layers.Conv2D() Examples

def build_cae_model(height=32, width=32, channel=3):
    """
    build convolutional autoencoder model
    """
    input_img = Input(shape=(height, width, channel))

    # encoder
    net = Conv2D(16, (3, 3), activation='relu', padding='same')(input_img)
    net = MaxPooling2D((2, 2), padding='same')(net)
    net = Conv2D(8, (3, 3), activation='relu', padding='same')(net)
    net = MaxPooling2D((2, 2), padding='same')(net)
    net = Conv2D(4, (3, 3), activation='relu', padding='same')(net)
    encoded = MaxPooling2D((2, 2), padding='same', name='enc')(net)

    # decoder
    net = Conv2D(4, (3, 3), activation='relu', padding='same')(encoded)
    net = UpSampling2D((2, 2))(net)
    net = Conv2D(8, (3, 3), activation='relu', padding='same')(net)
    net = UpSampling2D((2, 2))(net)
    net = Conv2D(16, (3, 3), activation='relu', padding='same')(net)
    net = UpSampling2D((2, 2))(net)
    decoded = Conv2D(channel, (3, 3), activation='sigmoid', padding='same')(net)

    return Model(input_img, decoded) 

def DCGAN_discriminator():
    nb_filters = 64
    nb_conv = int(np.floor(np.log(128) / np.log(2)))
    list_filters = [nb_filters * min(8, (2 ** i)) for i in range(nb_conv)]

    input_img = Input(shape=(128, 128, 3))
    x = Conv2D(list_filters[0], (3, 3), strides=(2, 2), name="disc_conv2d_1", padding="same")(input_img)
    x = BatchNormalization(axis=-1)(x)
    x = LeakyReLU(0.2)(x)
    # Next convs
    for i, f in enumerate(list_filters[1:]):
        name = "disc_conv2d_%s" % (i + 2)
        x = Conv2D(f, (3, 3), strides=(2, 2), name=name, padding="same")(x)
        x = BatchNormalization(axis=-1)(x)
        x = LeakyReLU(0.2)(x)

    x_flat = Flatten()(x)
    x_out = Dense(1, activation="sigmoid", name="disc_dense")(x_flat)
    discriminator_model = Model(inputs=input_img, outputs=[x_out])
    return discriminator_model 

def _conv2d_same(x, filters, prefix, stride=1, kernel_size=3, rate=1):
    # 计算padding的数量，hw是否需要收缩
    if stride == 1:
        return Conv2D(filters,
                      (kernel_size, kernel_size),
                      strides=(stride, stride),
                      padding='same', use_bias=False,
                      dilation_rate=(rate, rate),
                      name=prefix)(x)
    else:
        kernel_size_effective = kernel_size + (kernel_size - 1) * (rate - 1)
        pad_total = kernel_size_effective - 1
        pad_beg = pad_total // 2
        pad_end = pad_total - pad_beg
        x = ZeroPadding2D((pad_beg, pad_end))(x)
        return Conv2D(filters,
                      (kernel_size, kernel_size),
                      strides=(stride, stride),
                      padding='valid', use_bias=False,
                      dilation_rate=(rate, rate),
                      name=prefix)(x) 

def build_model(x_train, num_classes):
        # Reset default graph. Keras leaves old ops in the graph,
        # which are ignored for execution but clutter graph
        # visualization in TensorBoard.
        tf.reset_default_graph()

        inputs = KL.Input(shape=x_train.shape[1:], name="input_image")
        x = KL.Conv2D(32, (3, 3), activation='relu', padding="same",
                      name="conv1")(inputs)
        x = KL.Conv2D(64, (3, 3), activation='relu', padding="same",
                      name="conv2")(x)
        x = KL.MaxPooling2D(pool_size=(2, 2), name="pool1")(x)
        x = KL.Flatten(name="flat1")(x)
        x = KL.Dense(128, activation='relu', name="dense1")(x)
        x = KL.Dense(num_classes, activation='softmax', name="dense2")(x)

        return KM.Model(inputs, x, "digit_classifier_model")

    # Load MNIST Data 

def InceptionLayer(self, a, b, c, d):
        def func(x):
            x1 = Conv2D(a, (1, 1), padding='same', activation='relu')(x)
            
            x2 = Conv2D(b, (1, 1), padding='same', activation='relu')(x)
            x2 = Conv2D(b, (3, 3), padding='same', activation='relu')(x2)
            
            x3 = Conv2D(c, (1, 1), padding='same', activation='relu')(x)
            x3 = Conv2D(c, (3, 3), dilation_rate = 2, strides = 1, padding='same', activation='relu')(x3)
            
            x4 = Conv2D(d, (1, 1), padding='same', activation='relu')(x)
            x4 = Conv2D(d, (3, 3), dilation_rate = 3, strides = 1, padding='same', activation='relu')(x4)

            y = Concatenate(axis = -1)([x1, x2, x3, x4])
            
            return y
        return func 

def d_block(inp, fil, p = True):

    skip = Conv2D(fil, 1, padding = 'same', kernel_initializer = 'he_normal')(inp)

    out = Conv2D(filters = fil, kernel_size = 3, padding = 'same', kernel_initializer = 'he_normal')(inp)
    out = LeakyReLU(0.2)(out)

    out = Conv2D(filters = fil, kernel_size = 3, padding = 'same', kernel_initializer = 'he_normal')(out)
    out = LeakyReLU(0.2)(out)

    out = Conv2D(fil, 1, padding = 'same', kernel_initializer = 'he_normal')(out)

    out = add([out, skip])
    out = LeakyReLU(0.2)(out)

    if p:
        out = AveragePooling2D()(out)

    return out 

def _conv_block(inp, convs, skip=True):
  x = inp
  count = 0
  len_convs = len(convs)
  for conv in convs:
    if count == (len_convs - 2) and skip:
      skip_connection = x
    count += 1
    if conv['stride'] > 1: x = ZeroPadding2D(((1,0),(1,0)))(x) # peculiar padding as darknet prefer left and top
    x = Conv2D(conv['filter'],
           conv['kernel'],
           strides=conv['stride'],
           padding='valid' if conv['stride'] > 1 else 'same', # peculiar padding as darknet prefer left and top
           name='conv_' + str(conv['layer_idx']),
           use_bias=False if conv['bnorm'] else True)(x)
    if conv['bnorm']: x = BatchNormalization(epsilon=0.001, name='bnorm_' + str(conv['layer_idx']))(x)
    if conv['leaky']: x = LeakyReLU(alpha=0.1, name='leaky_' + str(conv['layer_idx']))(x)
  return add([skip_connection, x]) if skip else x


#SPP block uses three pooling layers of sizes [5, 9, 13] with strides one and all outputs together with the input are concatenated to be fed
  #to the FC block 

def g_block(inp, fil, u = True):

    if u:
        out = UpSampling2D(interpolation = 'bilinear')(inp)
    else:
        out = Activation('linear')(inp)

    skip = Conv2D(fil, 1, padding = 'same', kernel_initializer = 'he_normal')(out)

    out = Conv2D(filters = fil, kernel_size = 3, padding = 'same', kernel_initializer = 'he_normal')(out)
    out = LeakyReLU(0.2)(out)

    out = Conv2D(filters = fil, kernel_size = 3, padding = 'same', kernel_initializer = 'he_normal')(out)
    out = LeakyReLU(0.2)(out)

    out = Conv2D(fil, 1, padding = 'same', kernel_initializer = 'he_normal')(out)

    out = add([out, skip])
    out = LeakyReLU(0.2)(out)

    return out 

def _bn_relu_conv_block(input, filters, kernel=(3, 3), stride=(1, 1), weight_decay=5e-4):
    ''' Adds a Batchnorm-Relu-Conv block for DPN
    Args:
        input: input tensor
        filters: number of output filters
        kernel: convolution kernel size
        stride: stride of convolution
    Returns: a keras tensor
    '''
    channel_axis = 1 if K.image_data_format() == 'channels_first' else -1

    x = Conv2D(filters, kernel, padding='same', use_bias=False, kernel_initializer='he_normal',
               kernel_regularizer=l2(weight_decay), strides=stride)(input)
    x = BatchNormalization(axis=channel_axis)(x)
    x = Activation('relu')(x)
    return x 

def _initial_conv_block_inception(input, initial_conv_filters, weight_decay=5e-4):
    ''' Adds an initial conv block, with batch norm and relu for the DPN
    Args:
        input: input tensor
        initial_conv_filters: number of filters for initial conv block
        weight_decay: weight decay factor
    Returns: a keras tensor
    '''
    channel_axis = 1 if K.image_data_format() == 'channels_first' else -1

    x = Conv2D(initial_conv_filters, (7, 7), padding='same', use_bias=False, kernel_initializer='he_normal',
               kernel_regularizer=l2(weight_decay), strides=(2, 2))(input)
    x = BatchNormalization(axis=channel_axis)(x)
    x = Activation('relu')(x)

    x = MaxPooling2D((3, 3), strides=(2, 2), padding='same')(x)

    return x 

def conv2d(h_num_filters, h_filter_width, h_stride, h_use_bias):

    def compile_fn(di, dh):
        layer = layers.Conv2D(dh['num_filters'], (dh['filter_width'],) * 2,
                              strides=(dh['stride'],) * 2,
                              use_bias=dh['use_bias'],
                              padding='SAME')

        def fn(di):
            return {'out': layer(di['in'])}

        return fn

    return siso_keras_module(
        'Conv2D', compile_fn, {
            'num_filters': h_num_filters,
            'filter_width': h_filter_width,
            'stride': h_stride,
            'use_bias': h_use_bias,
        }) 

def conv_block(input_tensor, kernel_size, filters, stage, block, strides=(1, 1), dilation=1):
    """conv_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
    """
    filters1, filters2, filters3 = filters
    if K.image_data_format() == 'channels_last':
        bn_axis = 3
    else:
        bn_axis = 1
    conv_name_base = 'res' + str(stage) + block + '_branch'
    bn_name_base = 'bn' + str(stage) + block + '_branch'

    x = Conv2D(filters1, (1, 1), strides=strides, name=conv_name_base + '2a', use_bias=False)(input_tensor)
    x = BN(axis=bn_axis, name=bn_name_base + '2a')(x)
    x = Activation('relu')(x)

    x = Conv2D(filters2, kernel_size, padding='same', name=conv_name_base + '2b', use_bias=False, dilation_rate=dilation)(x)
    x = BN(axis=bn_axis, name=bn_name_base + '2b')(x)
    x = Activation('relu')(x)

    x = Conv2D(filters3, (1, 1), name=conv_name_base + '2c', use_bias=False)(x)
    x = BN(axis=bn_axis, name=bn_name_base + '2c')(x)

    shortcut = Conv2D(filters3, (1, 1), strides=strides, name=conv_name_base + '1', use_bias=False)(input_tensor)
    shortcut = BN(axis=bn_axis, name=bn_name_base + '1')(shortcut)

    x = layers.add([x, shortcut])
    x = Activation('relu')(x)
    return x 

def SepConv_BN(x, filters, prefix, stride=1, kernel_size=3, rate=1, depth_activation=False, epsilon=1e-3):
    # 计算padding的数量，hw是否需要收缩
    if stride == 1:
        depth_padding = 'same'
    else:
        kernel_size_effective = kernel_size + (kernel_size - 1) * (rate - 1)
        pad_total = kernel_size_effective - 1
        pad_beg = pad_total // 2
        pad_end = pad_total - pad_beg
        x = ZeroPadding2D((pad_beg, pad_end))(x)
        depth_padding = 'valid'
    
    # 如果需要激活函数
    if not depth_activation:
        x = Activation('relu')(x)

    # 分离卷积，首先3x3分离卷积，再1x1卷积
    # 3x3采用膨胀卷积
    x = DepthwiseConv2D((kernel_size, kernel_size), strides=(stride, stride), dilation_rate=(rate, rate),
                        padding=depth_padding, use_bias=False, name=prefix + '_depthwise')(x)
    x = BatchNormalization(name=prefix + '_depthwise_BN', epsilon=epsilon)(x)
    if depth_activation:
        x = Activation('relu')(x)

    # 1x1卷积，进行压缩
    x = Conv2D(filters, (1, 1), padding='same',
               use_bias=False, name=prefix + '_pointwise')(x)
    x = BatchNormalization(name=prefix + '_pointwise_BN', epsilon=epsilon)(x)
    if depth_activation:
        x = Activation('relu')(x)

    return x 

def _grouped_convolution_block(input, grouped_channels, cardinality, strides, weight_decay=5e-4):
    ''' Adds a grouped convolution block. It is an equivalent block from the paper
    Args:
        input: input tensor
        grouped_channels: grouped number of filters
        cardinality: cardinality factor describing the number of groups
        strides: performs strided convolution for downscaling if > 1
        weight_decay: weight decay term
    Returns: a keras tensor
    '''
    init = input
    channel_axis = 1 if K.image_data_format() == 'channels_first' else -1

    group_list = []

    if cardinality == 1:
        # with cardinality 1, it is a standard convolution
        x = Conv2D(grouped_channels, (3, 3), padding='same', use_bias=False, strides=strides,
                   kernel_initializer='he_normal', kernel_regularizer=l2(weight_decay))(init)
        x = BatchNormalization(axis=channel_axis)(x)
        x = Activation('relu')(x)
        return x

    for c in range(cardinality):
        x = Lambda(lambda z: z[:, :, :, c * grouped_channels:(c + 1) * grouped_channels]
                   if K.image_data_format() == 'channels_last' else
                   lambda z: z[:, c * grouped_channels:(c + 1) * grouped_channels, :, :])(input)

        x = Conv2D(grouped_channels, (3, 3), padding='same', use_bias=False, strides=strides,
                   kernel_initializer='he_normal', kernel_regularizer=l2(weight_decay))(x)

        group_list.append(x)

    group_merge = concatenate(group_list, axis=channel_axis)
    group_merge = BatchNormalization(axis=channel_axis)(group_merge)
    group_merge = Activation('relu')(group_merge)
    return group_merge 

def conv_block(input_tensor, kernel_size, filters, stage, block,
               strides=(2, 2), use_bias=True, train_bn=True):
    """conv_block is the block that has a conv layer at shortcut
    # Arguments
        input_tensor: input tensor
        kernel_size: default 3, the kernel size of middle conv layer at main path
        filters: list of integers, the nb_filters 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
        use_bias: Boolean. To use or not use a bias in conv layers.
        train_bn: Boolean. Train or freeze Batch Norm layers
    Note that from stage 3, the first conv layer at main path is with subsample=(2,2)
    And the shortcut should have subsample=(2,2) as well
    """
    nb_filter1, nb_filter2, nb_filter3 = filters
    conv_name_base = 'res' + str(stage) + block + '_branch'
    bn_name_base = 'bn' + str(stage) + block + '_branch'

    x = KL.Conv2D(nb_filter1, (1, 1), strides=strides,
                  name=conv_name_base + '2a', use_bias=use_bias)(input_tensor)
    x = BatchNorm(name=bn_name_base + '2a')(x, training=train_bn)
    x = KL.Activation('relu')(x)

    x = KL.Conv2D(nb_filter2, (kernel_size, kernel_size), padding='same',
                  name=conv_name_base + '2b', use_bias=use_bias)(x)
    x = BatchNorm(name=bn_name_base + '2b')(x, training=train_bn)
    x = KL.Activation('relu')(x)

    x = KL.Conv2D(nb_filter3, (1, 1), name=conv_name_base +
                  '2c', use_bias=use_bias)(x)
    x = BatchNorm(name=bn_name_base + '2c')(x, training=train_bn)

    shortcut = KL.Conv2D(nb_filter3, (1, 1), strides=strides,
                         name=conv_name_base + '1', use_bias=use_bias)(input_tensor)
    shortcut = BatchNorm(name=bn_name_base + '1')(shortcut, training=train_bn)

    x = KL.Add()([x, shortcut])
    x = KL.Activation('relu', name='res' + str(stage) + block + '_out')(x)
    return x 

def identity_block(input_tensor, kernel_size, filters, stage, block,
                   use_bias=True, train_bn=True):
    """The identity_block is the block that has no conv layer at shortcut
    # Arguments
        input_tensor: input tensor
        kernel_size: default 3, the kernel size of middle conv layer at main path
        filters: list of integers, the nb_filters 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
        use_bias: Boolean. To use or not use a bias in conv layers.
        train_bn: Boolean. Train or freeze Batch Norm layers
    """
    nb_filter1, nb_filter2, nb_filter3 = filters
    conv_name_base = 'res' + str(stage) + block + '_branch'
    bn_name_base = 'bn' + str(stage) + block + '_branch'

    x = KL.Conv2D(nb_filter1, (1, 1), name=conv_name_base + '2a',
                  use_bias=use_bias)(input_tensor)
    x = BatchNorm(name=bn_name_base + '2a')(x, training=train_bn)
    x = KL.Activation('relu')(x)

    x = KL.Conv2D(nb_filter2, (kernel_size, kernel_size), padding='same',
                  name=conv_name_base + '2b', use_bias=use_bias)(x)
    x = BatchNorm(name=bn_name_base + '2b')(x, training=train_bn)
    x = KL.Activation('relu')(x)

    x = KL.Conv2D(nb_filter3, (1, 1), name=conv_name_base + '2c',
                  use_bias=use_bias)(x)
    x = BatchNorm(name=bn_name_base + '2c')(x, training=train_bn)

    x = KL.Add()([x, input_tensor])
    x = KL.Activation('relu', name='res' + str(stage) + block + '_out')(x)
    return x 

def load_traffic_sign_model(base=32, dense=512, num_classes=43):
    input_shape = (32, 32, 3)
    model = Sequential()
    model.add(Conv2D(base, (3, 3), padding='same',
                     input_shape=input_shape,
                     activation='relu'))
    model.add(Conv2D(base, (3, 3), activation='relu'))

    model.add(MaxPooling2D(pool_size=(2, 2)))
    model.add(Dropout(0.2))

    model.add(Conv2D(base * 2, (3, 3), padding='same',
                     activation='relu'))
    model.add(Conv2D(base * 2, (3, 3), activation='relu'))
    model.add(MaxPooling2D(pool_size=(2, 2)))
    model.add(Dropout(0.2))

    model.add(Conv2D(base * 4, (3, 3), padding='same',
                     activation='relu'))
    model.add(Conv2D(base * 4, (3, 3), activation='relu'))
    model.add(MaxPooling2D(pool_size=(2, 2)))
    model.add(Dropout(0.2))

    model.add(Flatten())
    model.add(Dense(dense, activation='relu'))
    model.add(Dropout(0.5))
    model.add(Dense(num_classes, activation='softmax'))

    opt = keras.optimizers.adam(lr=0.001, decay=1 * 10e-5)
    model.compile(loss='categorical_crossentropy', optimizer=opt, metrics=['accuracy'])

    return model 

def identity_block(input_tensor, kernel_size, filters, stage, block):
    '''The 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 nb_filters 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
    '''
    eps = 1.1e-5
    nb_filter1, nb_filter2, nb_filter3 = filters
    conv_name_base = 'res' + str(stage) + block + '_branch'
    bn_name_base = 'bn' + str(stage) + block + '_branch'
    scale_name_base = 'scale' + str(stage) + block + '_branch'

    x = Conv2D(nb_filter1, (1, 1), name=conv_name_base + '2a', use_bias=False)(input_tensor)
    x = BatchNormalization(epsilon=eps, axis=bn_axis, name=bn_name_base + '2a')(x)
    x = Scale(axis=bn_axis, name=scale_name_base + '2a')(x)
    x = Activation('relu', name=conv_name_base + '2a_relu')(x)

    x = ZeroPadding2D((1, 1), name=conv_name_base + '2b_zeropadding')(x)
    x = Conv2D(nb_filter2, (kernel_size, kernel_size),
                      name=conv_name_base + '2b', use_bias=False)(x)
    x = BatchNormalization(epsilon=eps, axis=bn_axis, name=bn_name_base + '2b')(x)
    x = Scale(axis=bn_axis, name=scale_name_base + '2b')(x)
    x = Activation('relu', name=conv_name_base + '2b_relu')(x)

    x = Conv2D(nb_filter3, (1, 1), name=conv_name_base + '2c', use_bias=False)(x)
    x = BatchNormalization(epsilon=eps, axis=bn_axis, name=bn_name_base + '2c')(x)
    x = Scale(axis=bn_axis, name=scale_name_base + '2c')(x)

    x = add([x, input_tensor], name='res' + str(stage) + block)
    x = Activation('relu', name='res' + str(stage) + block + '_relu')(x)
    return x 

def conv_block(input_tensor, kernel_size, filters, stage, block, strides=(2, 2)):
    """
    conv_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 nb_filters 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
    Note that from stage 3, the first conv layer at main path is with subsample=(2,2)
    And the shortcut should have subsample=(2,2) as well
    """

    nb_filter1, nb_filter2, nb_filter3 = filters
    conv_name_base = 'res' + str(stage) + block + '_branch'
    bn_name_base = 'bn' + str(stage) + block + '_branch'

    x = Conv2D(nb_filter1, (1, 1), strides=strides,
               name=conv_name_base + '2a')(input_tensor)
    x = BatchNormalization(axis=bn_axis, name=bn_name_base + '2a')(x)
    x = Activation('relu')(x)

    x = Conv2D(nb_filter2, (kernel_size, kernel_size), padding='same',
               name=conv_name_base + '2b')(x)
    x = BatchNormalization(axis=bn_axis, name=bn_name_base + '2b')(x)
    x = Activation('relu')(x)

    x = Conv2D(nb_filter3, (1, 1), name=conv_name_base + '2c')(x)
    x = BatchNormalization(axis=bn_axis, name=bn_name_base + '2c')(x)

    shortcut = Conv2D(nb_filter3, (1, 1), strides=strides,
                      name=conv_name_base + '1')(input_tensor)
    shortcut = BatchNormalization(axis=bn_axis, name=bn_name_base + '1')(shortcut)

    x = add([x, shortcut])
    x = Activation('relu')(x)
    return x 

def identity_block(input_tensor, kernel_size, filters, stage, block):
    """
    The 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 nb_filters 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
    """

    nb_filter1, nb_filter2, nb_filter3 = filters
    conv_name_base = 'res' + str(stage) + block + '_branch'
    bn_name_base = 'bn' + str(stage) + block + '_branch'

    x = Conv2D(nb_filter1, (1, 1), name=conv_name_base + '2a')(input_tensor)
    x = BatchNormalization(axis=bn_axis, name=bn_name_base + '2a')(x)
    x = Activation('relu')(x)

    x = Conv2D(nb_filter2, (kernel_size, kernel_size),
               padding='same', name=conv_name_base + '2b')(x)
    x = BatchNormalization(axis=bn_axis, name=bn_name_base + '2b')(x)
    x = Activation('relu')(x)

    x = Conv2D(nb_filter3, (1, 1), name=conv_name_base + '2c')(x)
    x = BatchNormalization(axis=bn_axis, name=bn_name_base + '2c')(x)

    x = add([x, input_tensor])
    x = Activation('relu')(x)
    return x 

def identity_block(input_tensor, kernel_size, filters, stage, block):
    """The identity block is the block that has no conv layer at shortcut.
    # Arguments
        input_tensor: input tensor
        kernel_size: default 3, the kernel size of middle conv layer at main path
        filters: list of integers, the filters 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.
    """
    filters1, filters2, filters3 = filters
    if K.image_data_format() == 'channels_last':
        bn_axis = 3
    else:
        bn_axis = 1
    conv_name_base = 'res' + str(stage) + block + '_branch'
    bn_name_base = 'bn' + str(stage) + block + '_branch'

    x = Conv2D(filters1, (1, 1), name=conv_name_base +'2a',kernel_regularizer=l2(0.0001))(input_tensor)
    x = BatchNormalization(axis=bn_axis, name=bn_name_base + '2a')(x)
    x = Activation('relu')(x)

    x = Conv2D(filters2, kernel_size,
               padding='same', name=conv_name_base + '2b',kernel_regularizer=l2(0.0001))(x)
    x = BatchNormalization(axis=bn_axis, name=bn_name_base + '2b')(x)
    x = Activation('relu')(x)

    x = Conv2D(filters3, (1, 1), name=conv_name_base + '2c',kernel_regularizer=l2(0.0001))(x)
    x = BatchNormalization(axis=bn_axis, name=bn_name_base + '2c')(x)

    x = layers.add([x, input_tensor])
    x = Activation('relu',name = 'act'+str(stage)+block+'_branch')(x)
    return x 

def identity_block(input_tensor, kernel_size, filters, stage, block, dilation=1):
    """The 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
        dilation: dilation of the intermediate convolution

    # Returns
        Output tensor for the block.
    """
    filters1, filters2, filters3 = filters
    if K.image_data_format() == 'channels_last':
        bn_axis = 3
    else:
        bn_axis = 1
    conv_name_base = 'res' + str(stage) + block + '_branch'
    bn_name_base = 'bn' + str(stage) + block + '_branch'

    x = Conv2D(filters1, (1, 1), name=conv_name_base + '2a', use_bias=False)(input_tensor)
    x = BN(axis=bn_axis, name=bn_name_base + '2a')(x)
    x = Activation('relu')(x)

    x = Conv2D(filters2, kernel_size, padding='same', name=conv_name_base + '2b', use_bias=False, dilation_rate=dilation)(x)
    x = BN(axis=bn_axis, name=bn_name_base + '2b')(x)
    x = Activation('relu')(x)

    x = Conv2D(filters3, (1, 1), name=conv_name_base + '2c', use_bias=False)(x)
    x = BN(axis=bn_axis, name=bn_name_base + '2c')(x)

    x = layers.add([x, input_tensor])
    x = Activation('relu')(x)
    return x 

def build_pyramid_pooling_module(res, input_shape, nb_classes, sigmoid=False, output_size=None):
    """Build the Pyramid Pooling Module."""
    # ---PSPNet concat layers with Interpolation
    feature_map_size = tuple(int(ceil(input_dim / 8.0)) for input_dim in input_shape)
    if K.image_data_format() == 'channels_last':
        bn_axis = 3
    else:
        bn_axis = 1
    print("PSP module will interpolate to a final feature map size of %s" %
          (feature_map_size, ))

    interp_block1 = psp_block(res, 1, feature_map_size, input_shape)
    interp_block2 = psp_block(res, 2, feature_map_size, input_shape)
    interp_block3 = psp_block(res, 3, feature_map_size, input_shape)
    interp_block6 = psp_block(res, 6, feature_map_size, input_shape)

    # concat all these layers. resulted
    res = Concatenate()([interp_block1,
                         interp_block2,
                         interp_block3,
                         interp_block6,
                         res])
    x = Conv2D(512, (1, 1), strides=(1, 1), padding="same", name="class_psp_reduce_conv", use_bias=False)(res)
    x = resnet.BN(bn_axis, name="class_psp_reduce_bn")(x)
    x = Activation('relu')(x)

    x = Conv2D(nb_classes, (1, 1), strides=(1, 1), name="class_psp_final_conv")(x)

    if output_size:
        x = Upsampling(output_size)(x)

    if sigmoid:
        x = Activation('sigmoid')(x)
    return x 

def psp_block(prev_layer, level, feature_map_shape, input_shape):
    if input_shape == (512, 512):
        kernel_strides_map = {1: [64, 64],
                              2: [32, 32],
                              3: [22, 21],
                              6: [11, 9]}  # TODO: Level 6: Kernel correct, but stride not exactly the same as Pytorch
    else:
        raise ValueError("Pooling parameters for input shape " + input_shape + " are not defined.")

    if K.image_data_format() == 'channels_last':
        bn_axis = 3
    else:
        bn_axis = 1

    names = [
        "class_psp_" + str(level) + "_conv",
        "class_psp_" + str(level) + "_bn"
    ]
    kernel = (kernel_strides_map[level][0], kernel_strides_map[level][0])
    strides = (kernel_strides_map[level][1], kernel_strides_map[level][1])
    prev_layer = AveragePooling2D(kernel, strides=strides)(prev_layer)
    prev_layer = Conv2D(512, (1, 1), strides=(1, 1), name=names[0], use_bias=False)(prev_layer)
    prev_layer = resnet.BN(bn_axis, name=names[1])(prev_layer)
    prev_layer = Activation('relu')(prev_layer)
    prev_layer = Upsampling(feature_map_shape)(prev_layer)
    return prev_layer 

def SepConv_BN(x, filters, prefix, stride=1, kernel_size=3, rate=1, depth_activation=False, epsilon=1e-3):
    # 计算padding的数量，hw是否需要收缩
    if stride == 1:
        depth_padding = 'same'
    else:
        kernel_size_effective = kernel_size + (kernel_size - 1) * (rate - 1)
        pad_total = kernel_size_effective - 1
        pad_beg = pad_total // 2
        pad_end = pad_total - pad_beg
        x = ZeroPadding2D((pad_beg, pad_end))(x)
        depth_padding = 'valid'
    
    # 如果需要激活函数
    if not depth_activation:
        x = Activation('relu')(x)

    # 分离卷积，首先3x3分离卷积，再1x1卷积
    # 3x3采用膨胀卷积
    x = DepthwiseConv2D((kernel_size, kernel_size), strides=(stride, stride), dilation_rate=(rate, rate),
                        padding=depth_padding, use_bias=False, name=prefix + '_depthwise')(x)
    x = BatchNormalization(name=prefix + '_depthwise_BN', epsilon=epsilon)(x)
    if depth_activation:
        x = Activation('relu')(x)

    # 1x1卷积，进行压缩
    x = Conv2D(filters, (1, 1), padding='same',
               use_bias=False, name=prefix + '_pointwise')(x)
    x = BatchNormalization(name=prefix + '_pointwise_BN', epsilon=epsilon)(x)
    if depth_activation:
        x = Activation('relu')(x)

    return x 

def _inverted_res_block(inputs, expansion, stride, alpha, filters, block_id, skip_connection, rate=1):
    in_channels = inputs.shape[-1].value  # inputs._keras_shape[-1]
    pointwise_conv_filters = int(filters * alpha)
    pointwise_filters = _make_divisible(pointwise_conv_filters, 8)
    x = inputs
    prefix = 'expanded_conv_{}_'.format(block_id)
    if block_id:
        # Expand

        x = Conv2D(expansion * in_channels, kernel_size=1, padding='same',
                   use_bias=False, activation=None,
                   name=prefix + 'expand')(x)
        x = BatchNormalization(epsilon=1e-3, momentum=0.999,
                               name=prefix + 'expand_BN')(x)
        x = Activation(relu6, name=prefix + 'expand_relu')(x)
    else:
        prefix = 'expanded_conv_'
    # Depthwise
    x = DepthwiseConv2D(kernel_size=3, strides=stride, activation=None,
                        use_bias=False, padding='same', dilation_rate=(rate, rate),
                        name=prefix + 'depthwise')(x)
    x = BatchNormalization(epsilon=1e-3, momentum=0.999,
                           name=prefix + 'depthwise_BN')(x)

    x = Activation(relu6, name=prefix + 'depthwise_relu')(x)

    # Project
    x = Conv2D(pointwise_filters,
               kernel_size=1, padding='same', use_bias=False, activation=None,
               name=prefix + 'project')(x)
    x = BatchNormalization(epsilon=1e-3, momentum=0.999,
                           name=prefix + 'project_BN')(x)

    if skip_connection:
        return Add(name=prefix + 'add')([inputs, x])

    # if in_channels == pointwise_filters and stride == 1:
    #    return Add(name='res_connect_' + str(block_id))([inputs, x])

    return x 

def SepConv_BN(x, filters, prefix, stride=1, kernel_size=3, rate=1, depth_activation=False, epsilon=1e-3):
    # 计算padding的数量，hw是否需要收缩
    if stride == 1:
        depth_padding = 'same'
    else:
        kernel_size_effective = kernel_size + (kernel_size - 1) * (rate - 1)
        pad_total = kernel_size_effective - 1
        pad_beg = pad_total // 2
        pad_end = pad_total - pad_beg
        x = ZeroPadding2D((pad_beg, pad_end))(x)
        depth_padding = 'valid'
    
    # 如果需要激活函数
    if not depth_activation:
        x = Activation('relu')(x)

    # 分离卷积，首先3x3分离卷积，再1x1卷积
    # 3x3采用膨胀卷积
    x = DepthwiseConv2D((kernel_size, kernel_size), strides=(stride, stride), dilation_rate=(rate, rate),
                        padding=depth_padding, use_bias=False, name=prefix + '_depthwise')(x)
    x = BatchNormalization(name=prefix + '_depthwise_BN', epsilon=epsilon)(x)
    if depth_activation:
        x = Activation('relu')(x)

    # 1x1卷积，进行压缩
    x = Conv2D(filters, (1, 1), padding='same',
               use_bias=False, name=prefix + '_pointwise')(x)
    x = BatchNormalization(name=prefix + '_pointwise_BN', epsilon=epsilon)(x)
    if depth_activation:
        x = Activation('relu')(x)

    return x 

def SepConv_BN(x, filters, prefix, stride=1, kernel_size=3, rate=1, depth_activation=False, epsilon=1e-3):
    # 计算padding的数量，hw是否需要收缩
    if stride == 1:
        depth_padding = 'same'
    else:
        kernel_size_effective = kernel_size + (kernel_size - 1) * (rate - 1)
        pad_total = kernel_size_effective - 1
        pad_beg = pad_total // 2
        pad_end = pad_total - pad_beg
        x = ZeroPadding2D((pad_beg, pad_end))(x)
        depth_padding = 'valid'
    
    # 如果需要激活函数
    if not depth_activation:
        x = Activation('relu')(x)

    # 分离卷积，首先3x3分离卷积，再1x1卷积
    # 3x3采用膨胀卷积
    x = DepthwiseConv2D((kernel_size, kernel_size), strides=(stride, stride), dilation_rate=(rate, rate),
                        padding=depth_padding, use_bias=False, name=prefix + '_depthwise')(x)
    x = BatchNormalization(name=prefix + '_depthwise_BN', epsilon=epsilon)(x)
    if depth_activation:
        x = Activation('relu')(x)

    # 1x1卷积，进行压缩
    x = Conv2D(filters, (1, 1), padding='same',
               use_bias=False, name=prefix + '_pointwise')(x)
    x = BatchNormalization(name=prefix + '_pointwise_BN', epsilon=epsilon)(x)
    if depth_activation:
        x = Activation('relu')(x)

    return x 

def _inverted_res_block(inputs, expansion, stride, alpha, filters, block_id, skip_connection, rate=1):
    in_channels = inputs.shape[-1].value  # inputs._keras_shape[-1]
    pointwise_conv_filters = int(filters * alpha)
    pointwise_filters = _make_divisible(pointwise_conv_filters, 8)
    x = inputs
    prefix = 'expanded_conv_{}_'.format(block_id)
    if block_id:
        # Expand

        x = Conv2D(expansion * in_channels, kernel_size=1, padding='same',
                   use_bias=False, activation=None,
                   name=prefix + 'expand')(x)
        x = BatchNormalization(epsilon=1e-3, momentum=0.999,
                               name=prefix + 'expand_BN')(x)
        x = Activation(relu6, name=prefix + 'expand_relu')(x)
    else:
        prefix = 'expanded_conv_'
    # Depthwise
    x = DepthwiseConv2D(kernel_size=3, strides=stride, activation=None,
                        use_bias=False, padding='same', dilation_rate=(rate, rate),
                        name=prefix + 'depthwise')(x)
    x = BatchNormalization(epsilon=1e-3, momentum=0.999,
                           name=prefix + 'depthwise_BN')(x)

    x = Activation(relu6, name=prefix + 'depthwise_relu')(x)

    # Project
    x = Conv2D(pointwise_filters,
               kernel_size=1, padding='same', use_bias=False, activation=None,
               name=prefix + 'project')(x)
    x = BatchNormalization(epsilon=1e-3, momentum=0.999,
                           name=prefix + 'project_BN')(x)

    if skip_connection:
        return Add(name=prefix + 'add')([inputs, x])

    # if in_channels == pointwise_filters and stride == 1:
    #    return Add(name='res_connect_' + str(block_id))([inputs, x])

    return x 

def ctpn(base_features, num_anchors, rnn_units=128, fc_units=512):
    """
    ctpn网络
    :param base_features: (B,H,W,C)
    :param num_anchors: anchors个数
    :param rnn_units:
    :param fc_units:
    :return:
    """
    x = layers.Conv2D(512, kernel_size=(3, 3), padding='same', name='pre_fc')(base_features)  # [B,H,W,512]
    # 沿着宽度方式做rnn
    rnn_forward = layers.TimeDistributed(layers.GRU(rnn_units, return_sequences=True, kernel_initializer='he_normal'),
                                         name='gru_forward')(x)
    rnn_backward = layers.TimeDistributed(
        layers.GRU(rnn_units, return_sequences=True, kernel_initializer='he_normal', go_backwards=True),
        name='gru_backward')(x)

    rnn_output = layers.Concatenate(name='gru_concat')([rnn_forward, rnn_backward])  # (B,H,W,256)

    # conv实现fc
    fc_output = layers.Conv2D(fc_units, kernel_size=(1, 1), activation='relu', name='fc_output')(
        rnn_output)  # (B,H,W,512)

    # 分类
    class_logits = layers.Conv2D(2 * num_anchors, kernel_size=(1, 1), name='cls')(fc_output)
    class_logits = layers.Reshape(target_shape=(-1, 2), name='cls_reshape')(class_logits)
    # 中心点垂直坐标和高度回归
    predict_deltas = layers.Conv2D(2 * num_anchors, kernel_size=(1, 1), name='deltas')(fc_output)
    predict_deltas = layers.Reshape(target_shape=(-1, 2), name='deltas_reshape')(predict_deltas)
    # 侧边精调(只需要预测x偏移即可)
    predict_side_deltas = layers.Conv2D(num_anchors, kernel_size=(1, 1), name='side_deltas')(fc_output)
    predict_side_deltas = layers.Reshape(target_shape=(-1, 1), name='side_deltas_reshape')(
        predict_side_deltas)
    return class_logits, predict_deltas, predict_side_deltas