深度学习之残差网络

资料下载

链接：https://pan.baidu.com/s/1mTqblxzWcYIRF7_kk8MQQA
提取码：7x6w

资料的下载真的很感谢(14条消息) 【中文】【吴恩达课后编程作业】Course 4 – 卷积神经网络 – 第二周作业_何宽的博客-CSDN博客

我找了几天resnet50.h5。

【博主使用的python版本：3.6.8】

对于此作业，您将使用 Keras。

在进入问题之前，请运行下面的单元格以加载所需的包。

import tensorflow as tfimport numpy as npimport scipy.miscfrom tensorflow.keras.applications.resnet_v2 import ResNet50V2from tensorflow.keras.preprocessing import imagefrom tensorflow.keras.applications.resnet_v2 import preprocess_input, decode_predictionsfrom tensorflow.keras import layersfrom tensorflow.keras.layers import Input, Add, Dense, Activation, ZeroPadding2D, BatchNormalization, Flatten, Conv2D, AveragePooling2D, MaxPooling2D, GlobalMaxPooling2Dfrom tensorflow.keras.models import Model, load_modelfrom resnets_utils import *from tensorflow.keras.initializers import random_uniform, glorot_uniform, constant, identityfrom tensorflow.python.framework.ops import EagerTensorfrom matplotlib.pyplot import imshowfrom test_utils import summary, comparatorimport public_tests

– 非常深的神经网络的问题

非常深度网络的主要好处是它可以表示非常复杂的功能。它还可以学习许多不同抽象级别的特征，从边缘（在较浅的层，更接近输入）到非常复杂的特征（在更深的层，更接近输出）。
但是，使用更深的网络并不总是有帮助。训练梯度的一个巨大障碍是梯度消失：非常深的网络通常有一个梯度信号，很快就会归零，从而使梯度下降变得非常慢。
更具体地说，在梯度下降期间，当您从最后一层反向传播回第一层时，您将乘以每一步的权重矩阵，因此梯度可以呈指数级迅速减小到零（或者在极少数情况下，指数级快速增长并“爆炸”，因为获得非常大的值）。
因此，在训练过程中，您可能会看到随着训练的进行，较浅层的梯度的大小（或范数）会非常迅速地减小到零，如下所示：

构建一个残差网络

左图显示了通过网络的“主要路径”。右侧的图像将快捷方式添加到主路径。通过将这些 ResNet 块堆叠在一起，您可以形成一个非常深的网络。
讲座提到，使用带有快捷方式的 ResNet 块也使其中一个块学习恒等函数变得非常容易。这意味着您可以堆叠额外的 ResNet 块，而几乎没有损害训练集性能的风险。
在这一点上，还有一些证据表明，学习恒等函数的便利性解释了ResNets的卓越性能，甚至超过了跳过连接对梯度消失的帮助。
ResNet 中使用两种主要类型的块，主要取决于输入/输出尺寸是相同还是不同。您将实现它们：“标识块”和“卷积块”。

恒等块（Identity block）

恒等块是 ResNet 中使用的标准块，对应于输入激活（例如a^[L]）与输出激活（例如a^[L+!]）具有相同维度的情况。为了充实 ResNet 身份块中发生的不同步骤，下面是一个显示各个步骤的替代图表：

上面的路径是“捷径”。较低的路径是“主要路径”。在此图中，请注意每层中的 CONV2D 和 ReLU 步骤。为了加快训练速度，添加了 BatchNorm 步骤。不要担心这很难实现 – 你会看到BatchNorm只是Keras中的一行代码！

在本练习中，您将实际实现此标识块的一个功能稍微强大的版本，其中跳过连接“跳过”3 个隐藏层而不是 2 个层。它看起来像这样：

主路径的第一部分：

第一个 CONV2D 具有F1形状为（1，1）和步幅为（1，1）的过滤器。它的填充是“有效的”。使用 0 作为随机统一初始化的种子：kernel_initializer = initializer(seed=0).
第一个 BatchNorm 是规范化“channel”轴。
然后应用 ReLU 激活函数。这没有超参数。

主路径的第二部分：

第二个 CONV2D 具有F2形状(f,f)和步幅为（1，1）的过滤器。它的填充是“相同的”。使用 0 作为随机统一初始化的种子：kernel_initializer = initializer(seed=0)
第二个 BatchNorm 是规范化“channel”轴。
然后应用 ReLU 激活函数。这没有超参数。

主路径的第三部分：

第三个 CONV2D 具有F3形状（1，1）和步幅为（1，1）的过滤器。它的填充是“相同的”。使用 0 作为随机统一初始化的种子：kernel_initializer = initializer(seed=0)
第三个 BatchNorm 是规范化“channel”轴。
然后应用 ReLU 激活函数。这没有超参数。

主路径的最后一部分：

第 3 层 X 的X_shortcut和输出相加。
提示：语法看起来像 Add（）（[var1，var2]）
然后应用 ReLU 激活函数。这没有超参数。

接下来我们就要实现残差网络的恒等块了

我们已将初始值设定项参数添加到函数中。此参数接收一个初始值设定项函数，类似于包 tensorflow.keras.initializers 或任何其他自定义初始值设定项中包含的函数。默认情况下，它将设置为random_uniform

请记住，这些函数接受种子参数，该参数可以是所需的任何值，但在此笔记本中必须将其设置为 0 才能进行评分。

下面是实际使用函数式 API 的强大功能来创建快捷方式路径的地方：

def identity_block(X, f, filters, training=True, initializer=random_uniform):    """   实现图 4 中定义的恒等块        Arguments:    X -- 形状的输入张量（m、n_H_prev、n_W_prev、n_C_prev）    f -- 整数，指定主路径中间 CONV 窗口的形状    filters -- python 整数列表，定义主路径的 CONV 层中的过滤器数量    训练 -- True：在训练模式下行为                错误：在推理模式下行为    初始值设定项 -- 设置图层的初始权重。等于随机统一初始值设定项        Returns:    X -- output of the identity block, tensor of shape (n_H, n_W, n_C)    """        # Retrieve Filters    F1, F2, F3 = filters        # Save the input value. You'll need this later to add back to the main path.     X_shortcut = X    cache = []    # 主路径的第一个组成部分    X = Conv2D(filters = F1, kernel_size = 1, strides = (1,1), padding = 'valid', kernel_initializer = initializer(seed=0))(X)    X = BatchNormalization(axis = 3)(X, training = training) # Default axis    X = Activation('relu')(X)        ### START CODE HERE    ## Second component of main path (≈3 lines)    X = Conv2D(filters = F2, kernel_size = f,strides = (1, 1),padding='same',kernel_initializer = initializer(seed=0))(X)    X = BatchNormalization(axis = 3)(X, training=training)    X = Activation('relu')(X)    ## Third component of main path (≈2 lines)    X = Conv2D(filters = F3, kernel_size = 1, strides = (1, 1), padding='valid', kernel_initializer = initializer(seed=0))(X)    X = BatchNormalization(axis = 3)(X, training=training)        ## Final step: Add shortcut value to main path, and pass it through a RELU activation (≈2 lines)    X = Add()([X_shortcut,X])    X = Activation('relu')(X)    ### END CODE HERE    return X

我们来测试一下：

np.random.seed(1)X1 = np.ones((1, 4, 4, 3)) * -1X2 = np.ones((1, 4, 4, 3)) * 1X3 = np.ones((1, 4, 4, 3)) * 3#按着X1，X2，X3的顺序排序X = np.concatenate((X1, X2, X3), axis = 0).astype(np.float32)A3 = identity_block(X, f=2, filters=[4, 4, 3],                   initializer=lambda seed=0:constant(value=1),                   training=False)print('\033[1mWith training=False\033[0m\n')A3np = A3.numpy()print(np.around(A3.numpy()[:,(0,-1),:,:].mean(axis = 3), 5))resume = A3np[:,(0,-1),:,:].mean(axis = 3)print(resume[1, 1, 0])print('\n\033[1mWith training=True\033[0m\n')np.random.seed(1)A4 = identity_block(X, f=2, filters=[3, 3, 3],                   initializer=lambda seed=0:constant(value=1),                   training=True)print(np.around(A4.numpy()[:,(0,-1),:,:].mean(axis = 3), 5))public_tests.identity_block_test(identity_block)

With training=False[[[  0.        0.        0.        0.     ]  [  0.        0.        0.        0.     ]] [[192.71234 192.71234 192.71234  96.85617]  [ 96.85617  96.85617  96.85617  48.92808]] [[578.1371  578.1371  578.1371  290.5685 ]  [290.5685  290.5685  290.5685  146.78426]]]96.85617With training=True[[[0.      0.      0.      0.     ]  [0.      0.      0.      0.     ]] [[0.40739 0.40739 0.40739 0.40739]  [0.40739 0.40739 0.40739 0.40739]] [[4.99991 4.99991 4.99991 3.25948]  [3.25948 3.25948 3.25948 2.40739]]]All tests passed!

卷积块

ResNet“卷积块”是第二种块类型。当输入和输出维度不匹配时，可以使用这种类型的块。与标识块的区别在于快捷方式路径中有一个 CONV2D 层：

快捷路径中的 CONV2D 层用于将输入调整为不同的维度，以便尺寸在将快捷方式值添加回主路径所需的最终添加中匹配。（这与讲座中讨论的矩阵的作用类似。）
例如，要将激活维度的高度和宽度减少 2 倍，可以使用步幅为 2 的 1×1 卷积。
快捷方式路径上的 CONV2D 层不使用任何非线性激活函数。它的主要作用是仅应用一个（学习的）线性函数来减小输入的维度，以便维度与后面的加法步骤相匹配。
对于前面的练习，出于评分目的需要额外的初始值设定项参数，并且默认情况下已将其设置为 glorot_uniform

主路径的第一个组成部分：

第一个 CONV2D 具有F1个形状为（1，1）和步幅为（s，s）的过滤器。它的填充是“有效的”。使用 0 作为种子glorot_uniform kernel_initializer = 初始值设定项（seed=0）。
第一个 BatchNorm 是规范化“通道”轴。
然后应用 ReLU 激活函数。这没有超参数。

主路径的第二个组成部分：

第二个 CONV2D 具有F2个形状（f，f）和步幅为（1，1）的过滤器。它的填充是“相同的”。使用 0 作为种子glorot_uniform kernel_initializer = 初始值设定项（seed=0）。
第二个 BatchNorm 是规范化“通道”轴。
然后应用 ReLU 激活函数。这没有超参数。

主路径的第三个组成部分：

第三个 CONV2D 具有F3个形状为（1，1）和步幅为（1，1）的过滤器。它的填充是“有效的”。使用 0 作为种子glorot_uniform kernel_initializer = 初始值设定项（seed=0）。
第三个 BatchNorm 是规范化“通道”轴。请注意，此组件中没有 ReLU 激活函数。

快捷方式路径：

CONV2D 具有F3个形状为（1，1）和步幅为（s，s）的过滤器。它的填充是“有效的”。使用 0 作为种子glorot_uniform kernel_initializer = 初始值设定项（seed=0）。
BatchNorm正在规范化“通道”轴。

最后一步：

快捷方式和主路径值相加。
然后应用 ReLU 激活函数。这没有超参数。

def convolutional_block(X, f, filters, s = 2, training=True, initializer=glorot_uniform):    """    图 4 中定义的卷积块的实现        Arguments:    X -- 形状的输入张量（m、n_H_prev、n_W_prev、n_C_prev）    f -- 整数，指定主路径中间 CONV 窗口的形状    filters -- python 整数列表，定义主路径的 CONV 层中的过滤器数量    s -- 整数，指定要使用的步幅    训练 -- True：在训练模式下行为                错误：在推理模式下行为    初始值设定项 -- 设置图层的初始权重。等于 Glorot 统一初始值设定项，                   也称为泽维尔均匀初始值设定项。        Returns:    X -- output of the convolutional block, tensor of shape (n_H, n_W, n_C)    """        # Retrieve Filters    F1, F2, F3 = filters        # Save the input value    X_shortcut = X    ##### MAIN PATH #####        # First component of main path glorot_uniform(seed=0)    X = Conv2D(filters = F1, kernel_size = 1, strides = (s, s), padding='valid', kernel_initializer = initializer(seed=0))(X)    X = BatchNormalization(axis = 3)(X, training=training)    X = Activation('relu')(X)    ### START CODE HERE        ## Second component of main path (≈3 lines)    X = Conv2D(filters = F2, kernel_size = f,strides = (1, 1),padding='same',kernel_initializer = initializer(seed=0))(X)    X = BatchNormalization(axis = 3)(X, training=training)    X = Activation('relu')(X)    ## Third component of main path (≈2 lines)    X = Conv2D(filters = F3, kernel_size = 1, strides = (1, 1), padding='valid', kernel_initializer = initializer(seed=0))(X)    X = BatchNormalization(axis = 3)(X, training=training)        ##### SHORTCUT PATH ##### (≈2 lines)    X_shortcut = Conv2D(filters = F3, kernel_size = 1, strides = (s, s), padding='valid', kernel_initializer = initializer(seed=0))(X_shortcut)    X_shortcut = BatchNormalization(axis = 3)(X_shortcut, training=training)        ### END CODE HERE    # Final step: Add shortcut value to main path (Use this order [X, X_shortcut]), and pass it through a RELU activation    X = Add()([X, X_shortcut])    X = Activation('relu')(X)        return X

我们测试一下：

from outputs import convolutional_block_output1, convolutional_block_output2np.random.seed(1)#X = np.random.randn(3, 4, 4, 6).astype(np.float32)X1 = np.ones((1, 4, 4, 3)) * -1X2 = np.ones((1, 4, 4, 3)) * 1X3 = np.ones((1, 4, 4, 3)) * 3X = np.concatenate((X1, X2, X3), axis = 0).astype(np.float32)A = convolutional_block(X, f = 2, filters = [2, 4, 6], training=False)assert type(A) == EagerTensor, "Use only tensorflow and keras functions"assert tuple(tf.shape(A).numpy()) == (3, 2, 2, 6), "Wrong shape."assert np.allclose(A.numpy(), convolutional_block_output1), "Wrong values when training=False."print(A[0])B = convolutional_block(X, f = 2, filters = [2, 4, 6], training=True)assert np.allclose(B.numpy(), convolutional_block_output2), "Wrong values when training=True."print('\033[92mAll tests passed!')

tf.Tensor([[[0.         0.66683817 0.         0.         0.88853896 0.5274254 ]  [0.         0.65053666 0.         0.         0.89592844 0.49965227]] [[0.         0.6312079  0.         0.         0.8636247  0.47643146]  [0.         0.5688321  0.         0.         0.85534114 0.41709304]]], shape=(2, 2, 6), dtype=float32)All tests passed!

构建你的第一个残差网络（50层）

您现在拥有构建非常深入的 ResNet 所需的块。下图详细描述了该神经网络的架构。图中的“ID BLOCK”代表“身份块”，“ID BLOCK x3”表示您应该将 3 个恒等快块堆叠在一起。

零填充用（3，3）的填充填充输入

步骤一：

2D 卷积有 64 个形状为（7，7）的过滤器，并使用（2，2）的步幅。
BatchNorm 应用于输入的“通道”轴。
最大池化使用（3，3）窗口和（2，2）步幅。

步骤二：

卷积块使用三组大小为 [64，64，256] 的过滤器，“f”为 3，“s” 为 1。
2 个身份块使用三组大小为 [64，64，256] 的过滤器，“f”为 3。

步骤三:

卷积块使用三组大小为 [128，128，512] 的过滤器，“f”为 3，“s” 为 2。
3 个身份块使用三组大小为 [128，128，512] 的过滤器，“f”为 3。

步骤四：

卷积块使用三组大小为 [256， 256， 1024] 的滤波器，“f” 为 3，“s” 为 2。
这 5 个身份块使用三组大小为 [256、256、1024] 的过滤器，“f”为 3。

步骤五：

卷积块使用三组大小为 [512， 512， 2048] 的滤波器，“f”为 3，“s” 为 2。
2 个身份块使用三组大小为 [512， 512， 2048] 的过滤器，“f”为 3。

二维平均池化使用形状为（2，2）的窗口。
“扁平化”层没有任何超参数。
全连接（密集）层使用 softmax 激活将其输入减少到类数。

def ResNet50(input_shape = (64, 64, 3), classes = 6):    """    流行的 ResNet50 架构的分阶段实现：    CONV2D -> BATCHNORM -> RELU -> MAXPOOL -> CONVBLOCK -> IDBLOCK*2 -> CONVBLOCK -> IDBLOCK*3    -> CONVBLOCK -> IDBLOCK*5 -> CONVBLOCK -> IDBLOCK*2 -> AVGPOOL -> FLATTEN -> DENSE     Arguments:    input_shape -- shape of the images of the dataset    classes -- integer, number of classes    Returns:    model -- a Model() instance in Keras    """        # Define the input as a tensor with shape input_shape    X_input = Input(input_shape)        # Zero-Padding    X = ZeroPadding2D((3, 3))(X_input)        # Stage 1    X = Conv2D(64, (7, 7), strides = (2, 2), kernel_initializer = glorot_uniform(seed=0))(X)    X = BatchNormalization(axis = 3)(X)    X = Activation('relu')(X)    X = MaxPooling2D((3, 3), strides=(2, 2))(X)    # Stage 2    X = convolutional_block(X, f = 3, filters = [64, 64, 256], s = 1)    X = identity_block(X, 3, [64, 64, 256])    X = identity_block(X, 3, [64, 64, 256])    ### START CODE HERE        ## Stage 3 (≈4 lines)    X = convolutional_block(X, f = 3, filters = [128,128,512], s = 2)    X = identity_block(X, 3,  [128,128,512])    X = identity_block(X, 3,  [128,128,512])    X = identity_block(X, 3,  [128,128,512])        ## Stage 4 (≈6 lines)    X = convolutional_block(X, f = 3, filters = [256, 256, 1024], s = 2)    X = identity_block(X, 3, [256, 256, 1024])    X = identity_block(X, 3, [256, 256, 1024])    X = identity_block(X, 3, [256, 256, 1024])    X = identity_block(X, 3, [256, 256, 1024])    X = identity_block(X, 3, [256, 256, 1024])    ## Stage 5 (≈3 lines)    X = convolutional_block(X, f = 3, filters = [512, 512, 2048], s = 2)    X = identity_block(X, 3, [512, 512, 2048])    X = identity_block(X, 3, [512, 512, 2048])    ## AVGPOOL (≈1 line). Use "X = AveragePooling2D(...)(X)"    X = AveragePooling2D((2, 2))(X)        ### END CODE HERE    # output layer    X = Flatten()(X)    X = Dense(classes, activation='softmax', kernel_initializer = glorot_uniform(seed=0))(X)            # Create model    model = Model(inputs = X_input, outputs = X)    return model

模型建立完成了，我们来看一下参数

model = ResNet50(input_shape = (64, 64, 3), classes = 6)print(model.summary())

Model: "model"__________________________________________________________________________________________________Layer (type)                    Output Shape         Param #     Connected to                     ==================================================================================================input_1 (InputLayer)            [(None, 64, 64, 3)]  0                                            __________________________________________________________________________________________________zero_padding2d (ZeroPadding2D)  (None, 70, 70, 3)    0           input_1[0][0]                    __________________________________________________________________________________________________conv2d_20 (Conv2D)              (None, 32, 32, 64)   9472        zero_padding2d[0][0]             __________________________________________________________________________________________________batch_normalization_20 (BatchNo (None, 32, 32, 64)   256         conv2d_20[0][0]                  __________________________________________________________________________________________________activation_18 (Activation)      (None, 32, 32, 64)   0           batch_normalization_20[0][0]     __________________________________________________________________________________________________max_pooling2d (MaxPooling2D)    (None, 15, 15, 64)   0           activation_18[0][0]              __________________________________________________________________________________________________conv2d_21 (Conv2D)              (None, 15, 15, 64)   4160        max_pooling2d[0][0]              __________________________________________________________________________________________________batch_normalization_21 (BatchNo (None, 15, 15, 64)   256         conv2d_21[0][0]                  __________________________________________________________________________________________________activation_19 (Activation)      (None, 15, 15, 64)   0           batch_normalization_21[0][0]     __________________________________________________________________________________________________conv2d_22 (Conv2D)              (None, 15, 15, 64)   36928       activation_19[0][0]              __________________________________________________________________________________________________batch_normalization_22 (BatchNo (None, 15, 15, 64)   256         conv2d_22[0][0]                  __________________________________________________________________________________________________activation_20 (Activation)      (None, 15, 15, 64)   0           batch_normalization_22[0][0]     __________________________________________________________________________________________________conv2d_23 (Conv2D)              (None, 15, 15, 256)  16640       activation_20[0][0]              __________________________________________________________________________________________________conv2d_24 (Conv2D)              (None, 15, 15, 256)  16640       max_pooling2d[0][0]              __________________________________________________________________________________________________batch_normalization_23 (BatchNo (None, 15, 15, 256)  1024        conv2d_23[0][0]                  __________________________________________________________________________________________________batch_normalization_24 (BatchNo (None, 15, 15, 256)  1024        conv2d_24[0][0]                  __________________________________________________________________________________________________add_6 (Add)                     (None, 15, 15, 256)  0           batch_normalization_23[0][0]                                                                      batch_normalization_24[0][0]     __________________________________________________________________________________________________activation_21 (Activation)      (None, 15, 15, 256)  0           add_6[0][0]                      __________________________________________________________________________________________________conv2d_25 (Conv2D)              (None, 15, 15, 64)   16448       activation_21[0][0]              __________________________________________________________________________________________________batch_normalization_25 (BatchNo (None, 15, 15, 64)   256         conv2d_25[0][0]                  __________________________________________________________________________________________________activation_22 (Activation)      (None, 15, 15, 64)   0           batch_normalization_25[0][0]     __________________________________________________________________________________________________conv2d_26 (Conv2D)              (None, 15, 15, 64)   36928       activation_22[0][0]              __________________________________________________________________________________________________batch_normalization_26 (BatchNo (None, 15, 15, 64)   256         conv2d_26[0][0]                  __________________________________________________________________________________________________activation_23 (Activation)      (None, 15, 15, 64)   0           batch_normalization_26[0][0]     __________________________________________________________________________________________________conv2d_27 (Conv2D)              (None, 15, 15, 256)  16640       activation_23[0][0]              __________________________________________________________________________________________________batch_normalization_27 (BatchNo (None, 15, 15, 256)  1024        conv2d_27[0][0]                  __________________________________________________________________________________________________add_7 (Add)                     (None, 15, 15, 256)  0           activation_21[0][0]                                                                               batch_normalization_27[0][0]     __________________________________________________________________________________________________activation_24 (Activation)      (None, 15, 15, 256)  0           add_7[0][0]                      __________________________________________________________________________________________________conv2d_28 (Conv2D)              (None, 15, 15, 64)   16448       activation_24[0][0]              __________________________________________________________________________________________________batch_normalization_28 (BatchNo (None, 15, 15, 64)   256         conv2d_28[0][0]                  __________________________________________________________________________________________________activation_25 (Activation)      (None, 15, 15, 64)   0           batch_normalization_28[0][0]     __________________________________________________________________________________________________conv2d_29 (Conv2D)              (None, 15, 15, 64)   36928       activation_25[0][0]              __________________________________________________________________________________________________batch_normalization_29 (BatchNo (None, 15, 15, 64)   256         conv2d_29[0][0]                  __________________________________________________________________________________________________activation_26 (Activation)      (None, 15, 15, 64)   0           batch_normalization_29[0][0]     __________________________________________________________________________________________________conv2d_30 (Conv2D)              (None, 15, 15, 256)  16640       activation_26[0][0]              __________________________________________________________________________________________________batch_normalization_30 (BatchNo (None, 15, 15, 256)  1024        conv2d_30[0][0]                  __________________________________________________________________________________________________add_8 (Add)                     (None, 15, 15, 256)  0           activation_24[0][0]                                                                               batch_normalization_30[0][0]     __________________________________________________________________________________________________activation_27 (Activation)      (None, 15, 15, 256)  0           add_8[0][0]                      __________________________________________________________________________________________________conv2d_31 (Conv2D)              (None, 8, 8, 128)    32896       activation_27[0][0]              __________________________________________________________________________________________________batch_normalization_31 (BatchNo (None, 8, 8, 128)    512         conv2d_31[0][0]                  __________________________________________________________________________________________________activation_28 (Activation)      (None, 8, 8, 128)    0           batch_normalization_31[0][0]     __________________________________________________________________________________________________conv2d_32 (Conv2D)              (None, 8, 8, 128)    147584      activation_28[0][0]              __________________________________________________________________________________________________batch_normalization_32 (BatchNo (None, 8, 8, 128)    512         conv2d_32[0][0]                  __________________________________________________________________________________________________activation_29 (Activation)      (None, 8, 8, 128)    0           batch_normalization_32[0][0]     __________________________________________________________________________________________________conv2d_33 (Conv2D)              (None, 8, 8, 512)    66048       activation_29[0][0]              __________________________________________________________________________________________________conv2d_34 (Conv2D)              (None, 8, 8, 512)    131584      activation_27[0][0]              __________________________________________________________________________________________________batch_normalization_33 (BatchNo (None, 8, 8, 512)    2048        conv2d_33[0][0]                  __________________________________________________________________________________________________batch_normalization_34 (BatchNo (None, 8, 8, 512)    2048        conv2d_34[0][0]                  __________________________________________________________________________________________________add_9 (Add)                     (None, 8, 8, 512)    0           batch_normalization_33[0][0]                                                                      batch_normalization_34[0][0]     __________________________________________________________________________________________________activation_30 (Activation)      (None, 8, 8, 512)    0           add_9[0][0]                      __________________________________________________________________________________________________conv2d_35 (Conv2D)              (None, 8, 8, 128)    65664       activation_30[0][0]              __________________________________________________________________________________________________batch_normalization_35 (BatchNo (None, 8, 8, 128)    512         conv2d_35[0][0]                  __________________________________________________________________________________________________activation_31 (Activation)      (None, 8, 8, 128)    0           batch_normalization_35[0][0]     __________________________________________________________________________________________________conv2d_36 (Conv2D)              (None, 8, 8, 128)    147584      activation_31[0][0]              __________________________________________________________________________________________________batch_normalization_36 (BatchNo (None, 8, 8, 128)    512         conv2d_36[0][0]                  __________________________________________________________________________________________________activation_32 (Activation)      (None, 8, 8, 128)    0           batch_normalization_36[0][0]     __________________________________________________________________________________________________conv2d_37 (Conv2D)              (None, 8, 8, 512)    66048       activation_32[0][0]              __________________________________________________________________________________________________batch_normalization_37 (BatchNo (None, 8, 8, 512)    2048        conv2d_37[0][0]                  __________________________________________________________________________________________________add_10 (Add)                    (None, 8, 8, 512)    0           activation_30[0][0]                                                                               batch_normalization_37[0][0]     __________________________________________________________________________________________________activation_33 (Activation)      (None, 8, 8, 512)    0           add_10[0][0]                     __________________________________________________________________________________________________conv2d_38 (Conv2D)              (None, 8, 8, 128)    65664       activation_33[0][0]              __________________________________________________________________________________________________batch_normalization_38 (BatchNo (None, 8, 8, 128)    512         conv2d_38[0][0]                  __________________________________________________________________________________________________activation_34 (Activation)      (None, 8, 8, 128)    0           batch_normalization_38[0][0]     __________________________________________________________________________________________________conv2d_39 (Conv2D)              (None, 8, 8, 128)    147584      activation_34[0][0]              __________________________________________________________________________________________________batch_normalization_39 (BatchNo (None, 8, 8, 128)    512         conv2d_39[0][0]                  __________________________________________________________________________________________________activation_35 (Activation)      (None, 8, 8, 128)    0           batch_normalization_39[0][0]     __________________________________________________________________________________________________conv2d_40 (Conv2D)              (None, 8, 8, 512)    66048       activation_35[0][0]              __________________________________________________________________________________________________batch_normalization_40 (BatchNo (None, 8, 8, 512)    2048        conv2d_40[0][0]                  __________________________________________________________________________________________________add_11 (Add)                    (None, 8, 8, 512)    0           activation_33[0][0]                                                                               batch_normalization_40[0][0]     __________________________________________________________________________________________________activation_36 (Activation)      (None, 8, 8, 512)    0           add_11[0][0]                     __________________________________________________________________________________________________conv2d_41 (Conv2D)              (None, 8, 8, 128)    65664       activation_36[0][0]              __________________________________________________________________________________________________batch_normalization_41 (BatchNo (None, 8, 8, 128)    512         conv2d_41[0][0]                  __________________________________________________________________________________________________activation_37 (Activation)      (None, 8, 8, 128)    0           batch_normalization_41[0][0]     __________________________________________________________________________________________________conv2d_42 (Conv2D)              (None, 8, 8, 128)    147584      activation_37[0][0]              __________________________________________________________________________________________________batch_normalization_42 (BatchNo (None, 8, 8, 128)    512         conv2d_42[0][0]                  __________________________________________________________________________________________________activation_38 (Activation)      (None, 8, 8, 128)    0           batch_normalization_42[0][0]     __________________________________________________________________________________________________conv2d_43 (Conv2D)              (None, 8, 8, 512)    66048       activation_38[0][0]              __________________________________________________________________________________________________batch_normalization_43 (BatchNo (None, 8, 8, 512)    2048        conv2d_43[0][0]                  __________________________________________________________________________________________________add_12 (Add)                    (None, 8, 8, 512)    0           activation_36[0][0]                                                                               batch_normalization_43[0][0]     __________________________________________________________________________________________________activation_39 (Activation)      (None, 8, 8, 512)    0           add_12[0][0]                     __________________________________________________________________________________________________conv2d_44 (Conv2D)              (None, 4, 4, 256)    131328      activation_39[0][0]              __________________________________________________________________________________________________batch_normalization_44 (BatchNo (None, 4, 4, 256)    1024        conv2d_44[0][0]                  __________________________________________________________________________________________________activation_40 (Activation)      (None, 4, 4, 256)    0           batch_normalization_44[0][0]     __________________________________________________________________________________________________conv2d_45 (Conv2D)              (None, 4, 4, 256)    590080      activation_40[0][0]              __________________________________________________________________________________________________batch_normalization_45 (BatchNo (None, 4, 4, 256)    1024        conv2d_45[0][0]                  __________________________________________________________________________________________________activation_41 (Activation)      (None, 4, 4, 256)    0           batch_normalization_45[0][0]     __________________________________________________________________________________________________conv2d_46 (Conv2D)              (None, 4, 4, 1024)   263168      activation_41[0][0]              __________________________________________________________________________________________________conv2d_47 (Conv2D)              (None, 4, 4, 1024)   525312      activation_39[0][0]              __________________________________________________________________________________________________batch_normalization_46 (BatchNo (None, 4, 4, 1024)   4096        conv2d_46[0][0]                  __________________________________________________________________________________________________batch_normalization_47 (BatchNo (None, 4, 4, 1024)   4096        conv2d_47[0][0]                  __________________________________________________________________________________________________add_13 (Add)                    (None, 4, 4, 1024)   0           batch_normalization_46[0][0]                                                                      batch_normalization_47[0][0]     __________________________________________________________________________________________________activation_42 (Activation)      (None, 4, 4, 1024)   0           add_13[0][0]                     __________________________________________________________________________________________________conv2d_48 (Conv2D)              (None, 4, 4, 256)    262400      activation_42[0][0]              __________________________________________________________________________________________________batch_normalization_48 (BatchNo (None, 4, 4, 256)    1024        conv2d_48[0][0]                  __________________________________________________________________________________________________activation_43 (Activation)      (None, 4, 4, 256)    0           batch_normalization_48[0][0]     __________________________________________________________________________________________________conv2d_49 (Conv2D)              (None, 4, 4, 256)    590080      activation_43[0][0]              __________________________________________________________________________________________________batch_normalization_49 (BatchNo (None, 4, 4, 256)    1024        conv2d_49[0][0]                  __________________________________________________________________________________________________activation_44 (Activation)      (None, 4, 4, 256)    0           batch_normalization_49[0][0]     __________________________________________________________________________________________________conv2d_50 (Conv2D)              (None, 4, 4, 1024)   263168      activation_44[0][0]              __________________________________________________________________________________________________batch_normalization_50 (BatchNo (None, 4, 4, 1024)   4096        conv2d_50[0][0]                  __________________________________________________________________________________________________add_14 (Add)                    (None, 4, 4, 1024)   0           activation_42[0][0]                                                                               batch_normalization_50[0][0]     __________________________________________________________________________________________________activation_45 (Activation)      (None, 4, 4, 1024)   0           add_14[0][0]                     __________________________________________________________________________________________________conv2d_51 (Conv2D)              (None, 4, 4, 256)    262400      activation_45[0][0]              __________________________________________________________________________________________________batch_normalization_51 (BatchNo (None, 4, 4, 256)    1024        conv2d_51[0][0]                  __________________________________________________________________________________________________activation_46 (Activation)      (None, 4, 4, 256)    0           batch_normalization_51[0][0]     __________________________________________________________________________________________________conv2d_52 (Conv2D)              (None, 4, 4, 256)    590080      activation_46[0][0]              __________________________________________________________________________________________________batch_normalization_52 (BatchNo (None, 4, 4, 256)    1024        conv2d_52[0][0]                  __________________________________________________________________________________________________activation_47 (Activation)      (None, 4, 4, 256)    0           batch_normalization_52[0][0]     __________________________________________________________________________________________________conv2d_53 (Conv2D)              (None, 4, 4, 1024)   263168      activation_47[0][0]              __________________________________________________________________________________________________batch_normalization_53 (BatchNo (None, 4, 4, 1024)   4096        conv2d_53[0][0]                  __________________________________________________________________________________________________add_15 (Add)                    (None, 4, 4, 1024)   0           activation_45[0][0]                                                                               batch_normalization_53[0][0]     __________________________________________________________________________________________________activation_48 (Activation)      (None, 4, 4, 1024)   0           add_15[0][0]                     __________________________________________________________________________________________________conv2d_54 (Conv2D)              (None, 4, 4, 256)    262400      activation_48[0][0]              __________________________________________________________________________________________________batch_normalization_54 (BatchNo (None, 4, 4, 256)    1024        conv2d_54[0][0]                  __________________________________________________________________________________________________activation_49 (Activation)      (None, 4, 4, 256)    0           batch_normalization_54[0][0]     __________________________________________________________________________________________________conv2d_55 (Conv2D)              (None, 4, 4, 256)    590080      activation_49[0][0]              __________________________________________________________________________________________________batch_normalization_55 (BatchNo (None, 4, 4, 256)    1024        conv2d_55[0][0]                  __________________________________________________________________________________________________activation_50 (Activation)      (None, 4, 4, 256)    0           batch_normalization_55[0][0]     __________________________________________________________________________________________________conv2d_56 (Conv2D)              (None, 4, 4, 1024)   263168      activation_50[0][0]              __________________________________________________________________________________________________batch_normalization_56 (BatchNo (None, 4, 4, 1024)   4096        conv2d_56[0][0]                  __________________________________________________________________________________________________add_16 (Add)                    (None, 4, 4, 1024)   0           activation_48[0][0]                                                                               batch_normalization_56[0][0]     __________________________________________________________________________________________________activation_51 (Activation)      (None, 4, 4, 1024)   0           add_16[0][0]                     __________________________________________________________________________________________________conv2d_57 (Conv2D)              (None, 4, 4, 256)    262400      activation_51[0][0]              __________________________________________________________________________________________________batch_normalization_57 (BatchNo (None, 4, 4, 256)    1024        conv2d_57[0][0]                  __________________________________________________________________________________________________activation_52 (Activation)      (None, 4, 4, 256)    0           batch_normalization_57[0][0]     __________________________________________________________________________________________________conv2d_58 (Conv2D)              (None, 4, 4, 256)    590080      activation_52[0][0]              __________________________________________________________________________________________________batch_normalization_58 (BatchNo (None, 4, 4, 256)    1024        conv2d_58[0][0]                  __________________________________________________________________________________________________activation_53 (Activation)      (None, 4, 4, 256)    0           batch_normalization_58[0][0]     __________________________________________________________________________________________________conv2d_59 (Conv2D)              (None, 4, 4, 1024)   263168      activation_53[0][0]              __________________________________________________________________________________________________batch_normalization_59 (BatchNo (None, 4, 4, 1024)   4096        conv2d_59[0][0]                  __________________________________________________________________________________________________add_17 (Add)                    (None, 4, 4, 1024)   0           activation_51[0][0]                                                                               batch_normalization_59[0][0]     __________________________________________________________________________________________________activation_54 (Activation)      (None, 4, 4, 1024)   0           add_17[0][0]                     __________________________________________________________________________________________________conv2d_60 (Conv2D)              (None, 4, 4, 256)    262400      activation_54[0][0]              __________________________________________________________________________________________________batch_normalization_60 (BatchNo (None, 4, 4, 256)    1024        conv2d_60[0][0]                  __________________________________________________________________________________________________activation_55 (Activation)      (None, 4, 4, 256)    0           batch_normalization_60[0][0]     __________________________________________________________________________________________________conv2d_61 (Conv2D)              (None, 4, 4, 256)    590080      activation_55[0][0]              __________________________________________________________________________________________________batch_normalization_61 (BatchNo (None, 4, 4, 256)    1024        conv2d_61[0][0]                  __________________________________________________________________________________________________activation_56 (Activation)      (None, 4, 4, 256)    0           batch_normalization_61[0][0]     __________________________________________________________________________________________________conv2d_62 (Conv2D)              (None, 4, 4, 1024)   263168      activation_56[0][0]              __________________________________________________________________________________________________batch_normalization_62 (BatchNo (None, 4, 4, 1024)   4096        conv2d_62[0][0]                  __________________________________________________________________________________________________add_18 (Add)                    (None, 4, 4, 1024)   0           activation_54[0][0]                                                                               batch_normalization_62[0][0]     __________________________________________________________________________________________________activation_57 (Activation)      (None, 4, 4, 1024)   0           add_18[0][0]                     __________________________________________________________________________________________________conv2d_63 (Conv2D)              (None, 2, 2, 512)    524800      activation_57[0][0]              __________________________________________________________________________________________________batch_normalization_63 (BatchNo (None, 2, 2, 512)    2048        conv2d_63[0][0]                  __________________________________________________________________________________________________activation_58 (Activation)      (None, 2, 2, 512)    0           batch_normalization_63[0][0]     __________________________________________________________________________________________________conv2d_64 (Conv2D)              (None, 2, 2, 512)    2359808     activation_58[0][0]              __________________________________________________________________________________________________batch_normalization_64 (BatchNo (None, 2, 2, 512)    2048        conv2d_64[0][0]                  __________________________________________________________________________________________________activation_59 (Activation)      (None, 2, 2, 512)    0           batch_normalization_64[0][0]     __________________________________________________________________________________________________conv2d_65 (Conv2D)              (None, 2, 2, 2048)   1050624     activation_59[0][0]              __________________________________________________________________________________________________conv2d_66 (Conv2D)              (None, 2, 2, 2048)   2099200     activation_57[0][0]              __________________________________________________________________________________________________batch_normalization_65 (BatchNo (None, 2, 2, 2048)   8192        conv2d_65[0][0]                  __________________________________________________________________________________________________batch_normalization_66 (BatchNo (None, 2, 2, 2048)   8192        conv2d_66[0][0]                  __________________________________________________________________________________________________add_19 (Add)                    (None, 2, 2, 2048)   0           batch_normalization_65[0][0]                                                                      batch_normalization_66[0][0]     __________________________________________________________________________________________________activation_60 (Activation)      (None, 2, 2, 2048)   0           add_19[0][0]                     __________________________________________________________________________________________________conv2d_67 (Conv2D)              (None, 2, 2, 512)    1049088     activation_60[0][0]              __________________________________________________________________________________________________batch_normalization_67 (BatchNo (None, 2, 2, 512)    2048        conv2d_67[0][0]                  __________________________________________________________________________________________________activation_61 (Activation)      (None, 2, 2, 512)    0           batch_normalization_67[0][0]     __________________________________________________________________________________________________conv2d_68 (Conv2D)              (None, 2, 2, 512)    2359808     activation_61[0][0]              __________________________________________________________________________________________________batch_normalization_68 (BatchNo (None, 2, 2, 512)    2048        conv2d_68[0][0]                  __________________________________________________________________________________________________activation_62 (Activation)      (None, 2, 2, 512)    0           batch_normalization_68[0][0]     __________________________________________________________________________________________________conv2d_69 (Conv2D)              (None, 2, 2, 2048)   1050624     activation_62[0][0]              __________________________________________________________________________________________________batch_normalization_69 (BatchNo (None, 2, 2, 2048)   8192        conv2d_69[0][0]                  __________________________________________________________________________________________________add_20 (Add)                    (None, 2, 2, 2048)   0           activation_60[0][0]                                                                               batch_normalization_69[0][0]     __________________________________________________________________________________________________activation_63 (Activation)      (None, 2, 2, 2048)   0           add_20[0][0]                     __________________________________________________________________________________________________conv2d_70 (Conv2D)              (None, 2, 2, 512)    1049088     activation_63[0][0]              __________________________________________________________________________________________________batch_normalization_70 (BatchNo (None, 2, 2, 512)    2048        conv2d_70[0][0]                  __________________________________________________________________________________________________activation_64 (Activation)      (None, 2, 2, 512)    0           batch_normalization_70[0][0]     __________________________________________________________________________________________________conv2d_71 (Conv2D)              (None, 2, 2, 512)    2359808     activation_64[0][0]              __________________________________________________________________________________________________batch_normalization_71 (BatchNo (None, 2, 2, 512)    2048        conv2d_71[0][0]                  __________________________________________________________________________________________________activation_65 (Activation)      (None, 2, 2, 512)    0           batch_normalization_71[0][0]     __________________________________________________________________________________________________conv2d_72 (Conv2D)              (None, 2, 2, 2048)   1050624     activation_65[0][0]              __________________________________________________________________________________________________batch_normalization_72 (BatchNo (None, 2, 2, 2048)   8192        conv2d_72[0][0]                  __________________________________________________________________________________________________add_21 (Add)                    (None, 2, 2, 2048)   0           activation_63[0][0]                                                                               batch_normalization_72[0][0]     __________________________________________________________________________________________________activation_66 (Activation)      (None, 2, 2, 2048)   0           add_21[0][0]                     __________________________________________________________________________________________________average_pooling2d (AveragePooli (None, 1, 1, 2048)   0           activation_66[0][0]              __________________________________________________________________________________________________flatten (Flatten)               (None, 2048)         0           average_pooling2d[0][0]          __________________________________________________________________________________________________dense (Dense)                   (None, 6)            12294       flatten[0][0]                    ==================================================================================================Total params: 23,600,006Trainable params: 23,546,886Non-trainable params: 53,120__________________________________________________________________________________________________None

from outputs import ResNet50_summarymodel = ResNet50(input_shape = (64, 64, 3), classes = 6)

实体化结束，下面我们对模型进行编译

model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])

接下来我们就是加载数据集并进行训练

X_train_orig, Y_train_orig, X_test_orig, Y_test_orig, classes = load_dataset()# Normalize image vectorsX_train = X_train_orig / 255.X_test = X_test_orig / 255.# Convert training and test labels to one hot matricesY_train = convert_to_one_hot(Y_train_orig, 6).TY_test = convert_to_one_hot(Y_test_orig, 6).Tprint ("number of training examples = " + str(X_train.shape[0]))print ("number of test examples = " + str(X_test.shape[0]))print ("X_train shape: " + str(X_train.shape))print ("Y_train shape: " + str(Y_train.shape))print ("X_test shape: " + str(X_test.shape))print ("Y_test shape: " + str(Y_test.shape))

number of training examples = 1080number of test examples = 120X_train shape: (1080, 64, 64, 3)Y_train shape: (1080, 6)X_test shape: (120, 64, 64, 3)Y_test shape: (120, 6)

model.fit(X_train, Y_train, epochs = 10, batch_size = 32)

Epoch 1/1034/34 [==============================] - 10s 129ms/step - loss: 2.2749 - accuracy: 0.4417Epoch 2/1034/34 [==============================] - 3s 92ms/step - loss: 1.0135 - accuracy: 0.6889Epoch 3/1034/34 [==============================] - 3s 92ms/step - loss: 0.3830 - accuracy: 0.8694Epoch 4/1034/34 [==============================] - 3s 91ms/step - loss: 0.2390 - accuracy: 0.9241Epoch 5/1034/34 [==============================] - 3s 91ms/step - loss: 0.1527 - accuracy: 0.9519 0s - loss: 0.1Epoch 6/1034/34 [==============================] - 3s 91ms/step - loss: 0.1050 - accuracy: 0.9648Epoch 7/1034/34 [==============================] - 3s 91ms/step - loss: 0.1824 - accuracy: 0.9444Epoch 8/1034/34 [==============================] - 3s 92ms/step - loss: 0.5906 - accuracy: 0.8074Epoch 9/1034/34 [==============================] - 3s 91ms/step - loss: 0.4195 - accuracy: 0.8676Epoch 10/1034/34 [==============================] - 3s 91ms/step - loss: 0.3377 - accuracy: 0.9194
我们对模型进行评估

preds = model.evaluate(X_test, Y_test)print ("Loss = " + str(preds[0]))print ("Test Accuracy = " + str(preds[1]))

4/4 [==============================] - 1s 31ms/step - loss: 0.4844 - accuracy: 0.8417Loss = 0.4843602478504181Test Accuracy = 0.8416666388511658
博主已经在手势数据集上训练了自己的RESNET50模型的权重，你可以使用下面的代码载并运行博主的训练模型，

pre_trained_model = tf.keras.models.load_model('resnet50.h5')

然后测试一下博主训练出来的权值：

preds = pre_trained_model.evaluate(X_test, Y_test)print ("Loss = " + str(preds[0]))print ("Test Accuracy = " + str(preds[1]))

使用自己的图片做测试

img_path = 'C:/Users/Style/Desktop/kun.png'img = image.load_img(img_path, target_size=(64, 64))x = image.img_to_array(img)x = np.expand_dims(x, axis=0)x = x/255.0print('Input image shape:', x.shape)imshow(img)prediction = model.predict(x)print("Class prediction vector [p(0), p(1), p(2), p(3), p(4), p(5)] = ", prediction)print("Class:", np.argmax(prediction))

Input image shape: (1, 64, 64, 3)Class prediction vector [p(0), p(1), p(2), p(3), p(4), p(5)] =  [[0.01301673 0.8742848  0.00662233 0.05449386 0.02079306 0.03078919]]Class: 1

pre_trained_model.summary()

Model: "ResNet50"__________________________________________________________________________________________________Layer (type)                    Output Shape         Param #     Connected to                     ==================================================================================================input_1 (InputLayer)            [(None, 64, 64, 3)]  0                                            __________________________________________________________________________________________________zero_padding2d_1 (ZeroPadding2D (None, 70, 70, 3)    0           input_1[0][0]                    __________________________________________________________________________________________________conv1 (Conv2D)                  (None, 32, 32, 64)   9472        zero_padding2d_1[0][0]           __________________________________________________________________________________________________bn_conv1 (BatchNormalization)   (None, 32, 32, 64)   256         conv1[0][0]                      __________________________________________________________________________________________________activation_1 (Activation)       (None, 32, 32, 64)   0           bn_conv1[0][0]                   __________________________________________________________________________________________________max_pooling2d_1 (MaxPooling2D)  (None, 15, 15, 64)   0           activation_1[0][0]               __________________________________________________________________________________________________res2a_branch2a (Conv2D)         (None, 15, 15, 64)   4160        max_pooling2d_1[0][0]            __________________________________________________________________________________________________bn2a_branch2a (BatchNormalizati (None, 15, 15, 64)   256         res2a_branch2a[0][0]             __________________________________________________________________________________________________activation_2 (Activation)       (None, 15, 15, 64)   0           bn2a_branch2a[0][0]              __________________________________________________________________________________________________res2a_branch2b (Conv2D)         (None, 15, 15, 64)   36928       activation_2[0][0]               __________________________________________________________________________________________________bn2a_branch2b (BatchNormalizati (None, 15, 15, 64)   256         res2a_branch2b[0][0]             __________________________________________________________________________________________________activation_3 (Activation)       (None, 15, 15, 64)   0           bn2a_branch2b[0][0]              __________________________________________________________________________________________________res2a_branch2c (Conv2D)         (None, 15, 15, 256)  16640       activation_3[0][0]               __________________________________________________________________________________________________res2a_branch1 (Conv2D)          (None, 15, 15, 256)  16640       max_pooling2d_1[0][0]            __________________________________________________________________________________________________bn2a_branch2c (BatchNormalizati (None, 15, 15, 256)  1024        res2a_branch2c[0][0]             __________________________________________________________________________________________________bn2a_branch1 (BatchNormalizatio (None, 15, 15, 256)  1024        res2a_branch1[0][0]              __________________________________________________________________________________________________add_1 (Add)                     (None, 15, 15, 256)  0           bn2a_branch2c[0][0]                                                                               bn2a_branch1[0][0]               __________________________________________________________________________________________________activation_4 (Activation)       (None, 15, 15, 256)  0           add_1[0][0]                      __________________________________________________________________________________________________res2b_branch2a (Conv2D)         (None, 15, 15, 64)   16448       activation_4[0][0]               __________________________________________________________________________________________________bn2b_branch2a (BatchNormalizati (None, 15, 15, 64)   256         res2b_branch2a[0][0]             __________________________________________________________________________________________________activation_5 (Activation)       (None, 15, 15, 64)   0           bn2b_branch2a[0][0]              __________________________________________________________________________________________________res2b_branch2b (Conv2D)         (None, 15, 15, 64)   36928       activation_5[0][0]               __________________________________________________________________________________________________bn2b_branch2b (BatchNormalizati (None, 15, 15, 64)   256         res2b_branch2b[0][0]             __________________________________________________________________________________________________activation_6 (Activation)       (None, 15, 15, 64)   0           bn2b_branch2b[0][0]              __________________________________________________________________________________________________res2b_branch2c (Conv2D)         (None, 15, 15, 256)  16640       activation_6[0][0]               __________________________________________________________________________________________________bn2b_branch2c (BatchNormalizati (None, 15, 15, 256)  1024        res2b_branch2c[0][0]             __________________________________________________________________________________________________add_2 (Add)                     (None, 15, 15, 256)  0           bn2b_branch2c[0][0]                                                                               activation_4[0][0]               __________________________________________________________________________________________________activation_7 (Activation)       (None, 15, 15, 256)  0           add_2[0][0]                      __________________________________________________________________________________________________res2c_branch2a (Conv2D)         (None, 15, 15, 64)   16448       activation_7[0][0]               __________________________________________________________________________________________________bn2c_branch2a (BatchNormalizati (None, 15, 15, 64)   256         res2c_branch2a[0][0]             __________________________________________________________________________________________________activation_8 (Activation)       (None, 15, 15, 64)   0           bn2c_branch2a[0][0]              __________________________________________________________________________________________________res2c_branch2b (Conv2D)         (None, 15, 15, 64)   36928       activation_8[0][0]               __________________________________________________________________________________________________bn2c_branch2b (BatchNormalizati (None, 15, 15, 64)   256         res2c_branch2b[0][0]             __________________________________________________________________________________________________activation_9 (Activation)       (None, 15, 15, 64)   0           bn2c_branch2b[0][0]              __________________________________________________________________________________________________res2c_branch2c (Conv2D)         (None, 15, 15, 256)  16640       activation_9[0][0]               __________________________________________________________________________________________________bn2c_branch2c (BatchNormalizati (None, 15, 15, 256)  1024        res2c_branch2c[0][0]             __________________________________________________________________________________________________add_3 (Add)                     (None, 15, 15, 256)  0           bn2c_branch2c[0][0]                                                                               activation_7[0][0]               __________________________________________________________________________________________________activation_10 (Activation)      (None, 15, 15, 256)  0           add_3[0][0]                      __________________________________________________________________________________________________res3a_branch2a (Conv2D)         (None, 8, 8, 128)    32896       activation_10[0][0]              __________________________________________________________________________________________________bn3a_branch2a (BatchNormalizati (None, 8, 8, 128)    512         res3a_branch2a[0][0]             __________________________________________________________________________________________________activation_11 (Activation)      (None, 8, 8, 128)    0           bn3a_branch2a[0][0]              __________________________________________________________________________________________________res3a_branch2b (Conv2D)         (None, 8, 8, 128)    147584      activation_11[0][0]              __________________________________________________________________________________________________bn3a_branch2b (BatchNormalizati (None, 8, 8, 128)    512         res3a_branch2b[0][0]             __________________________________________________________________________________________________activation_12 (Activation)      (None, 8, 8, 128)    0           bn3a_branch2b[0][0]              __________________________________________________________________________________________________res3a_branch2c (Conv2D)         (None, 8, 8, 512)    66048       activation_12[0][0]              __________________________________________________________________________________________________res3a_branch1 (Conv2D)          (None, 8, 8, 512)    131584      activation_10[0][0]              __________________________________________________________________________________________________bn3a_branch2c (BatchNormalizati (None, 8, 8, 512)    2048        res3a_branch2c[0][0]             __________________________________________________________________________________________________bn3a_branch1 (BatchNormalizatio (None, 8, 8, 512)    2048        res3a_branch1[0][0]              __________________________________________________________________________________________________add_4 (Add)                     (None, 8, 8, 512)    0           bn3a_branch2c[0][0]                                                                               bn3a_branch1[0][0]               __________________________________________________________________________________________________activation_13 (Activation)      (None, 8, 8, 512)    0           add_4[0][0]                      __________________________________________________________________________________________________res3b_branch2a (Conv2D)         (None, 8, 8, 128)    65664       activation_13[0][0]              __________________________________________________________________________________________________bn3b_branch2a (BatchNormalizati (None, 8, 8, 128)    512         res3b_branch2a[0][0]             __________________________________________________________________________________________________activation_14 (Activation)      (None, 8, 8, 128)    0           bn3b_branch2a[0][0]              __________________________________________________________________________________________________res3b_branch2b (Conv2D)         (None, 8, 8, 128)    147584      activation_14[0][0]              __________________________________________________________________________________________________bn3b_branch2b (BatchNormalizati (None, 8, 8, 128)    512         res3b_branch2b[0][0]             __________________________________________________________________________________________________activation_15 (Activation)      (None, 8, 8, 128)    0           bn3b_branch2b[0][0]              __________________________________________________________________________________________________res3b_branch2c (Conv2D)         (None, 8, 8, 512)    66048       activation_15[0][0]              __________________________________________________________________________________________________bn3b_branch2c (BatchNormalizati (None, 8, 8, 512)    2048        res3b_branch2c[0][0]             __________________________________________________________________________________________________add_5 (Add)                     (None, 8, 8, 512)    0           bn3b_branch2c[0][0]                                                                               activation_13[0][0]              __________________________________________________________________________________________________activation_16 (Activation)      (None, 8, 8, 512)    0           add_5[0][0]                      __________________________________________________________________________________________________res3c_branch2a (Conv2D)         (None, 8, 8, 128)    65664       activation_16[0][0]              __________________________________________________________________________________________________bn3c_branch2a (BatchNormalizati (None, 8, 8, 128)    512         res3c_branch2a[0][0]             __________________________________________________________________________________________________activation_17 (Activation)      (None, 8, 8, 128)    0           bn3c_branch2a[0][0]              __________________________________________________________________________________________________res3c_branch2b (Conv2D)         (None, 8, 8, 128)    147584      activation_17[0][0]              __________________________________________________________________________________________________bn3c_branch2b (BatchNormalizati (None, 8, 8, 128)    512         res3c_branch2b[0][0]             __________________________________________________________________________________________________activation_18 (Activation)      (None, 8, 8, 128)    0           bn3c_branch2b[0][0]              __________________________________________________________________________________________________res3c_branch2c (Conv2D)         (None, 8, 8, 512)    66048       activation_18[0][0]              __________________________________________________________________________________________________bn3c_branch2c (BatchNormalizati (None, 8, 8, 512)    2048        res3c_branch2c[0][0]             __________________________________________________________________________________________________add_6 (Add)                     (None, 8, 8, 512)    0           bn3c_branch2c[0][0]                                                                               activation_16[0][0]              __________________________________________________________________________________________________activation_19 (Activation)      (None, 8, 8, 512)    0           add_6[0][0]                      __________________________________________________________________________________________________res3d_branch2a (Conv2D)         (None, 8, 8, 128)    65664       activation_19[0][0]              __________________________________________________________________________________________________bn3d_branch2a (BatchNormalizati (None, 8, 8, 128)    512         res3d_branch2a[0][0]             __________________________________________________________________________________________________activation_20 (Activation)      (None, 8, 8, 128)    0           bn3d_branch2a[0][0]              __________________________________________________________________________________________________res3d_branch2b (Conv2D)         (None, 8, 8, 128)    147584      activation_20[0][0]              __________________________________________________________________________________________________bn3d_branch2b (BatchNormalizati (None, 8, 8, 128)    512         res3d_branch2b[0][0]             __________________________________________________________________________________________________activation_21 (Activation)      (None, 8, 8, 128)    0           bn3d_branch2b[0][0]              __________________________________________________________________________________________________res3d_branch2c (Conv2D)         (None, 8, 8, 512)    66048       activation_21[0][0]              __________________________________________________________________________________________________bn3d_branch2c (BatchNormalizati (None, 8, 8, 512)    2048        res3d_branch2c[0][0]             __________________________________________________________________________________________________add_7 (Add)                     (None, 8, 8, 512)    0           bn3d_branch2c[0][0]                                                                               activation_19[0][0]              __________________________________________________________________________________________________activation_22 (Activation)      (None, 8, 8, 512)    0           add_7[0][0]                      __________________________________________________________________________________________________res4a_branch2a (Conv2D)         (None, 4, 4, 256)    131328      activation_22[0][0]              __________________________________________________________________________________________________bn4a_branch2a (BatchNormalizati (None, 4, 4, 256)    1024        res4a_branch2a[0][0]             __________________________________________________________________________________________________activation_23 (Activation)      (None, 4, 4, 256)    0           bn4a_branch2a[0][0]              __________________________________________________________________________________________________res4a_branch2b (Conv2D)         (None, 4, 4, 256)    590080      activation_23[0][0]              __________________________________________________________________________________________________bn4a_branch2b (BatchNormalizati (None, 4, 4, 256)    1024        res4a_branch2b[0][0]             __________________________________________________________________________________________________activation_24 (Activation)      (None, 4, 4, 256)    0           bn4a_branch2b[0][0]              __________________________________________________________________________________________________res4a_branch2c (Conv2D)         (None, 4, 4, 1024)   263168      activation_24[0][0]              __________________________________________________________________________________________________res4a_branch1 (Conv2D)          (None, 4, 4, 1024)   525312      activation_22[0][0]              __________________________________________________________________________________________________bn4a_branch2c (BatchNormalizati (None, 4, 4, 1024)   4096        res4a_branch2c[0][0]             __________________________________________________________________________________________________bn4a_branch1 (BatchNormalizatio (None, 4, 4, 1024)   4096        res4a_branch1[0][0]              __________________________________________________________________________________________________add_8 (Add)                     (None, 4, 4, 1024)   0           bn4a_branch2c[0][0]                                                                               bn4a_branch1[0][0]               __________________________________________________________________________________________________activation_25 (Activation)      (None, 4, 4, 1024)   0           add_8[0][0]                      __________________________________________________________________________________________________res4b_branch2a (Conv2D)         (None, 4, 4, 256)    262400      activation_25[0][0]              __________________________________________________________________________________________________bn4b_branch2a (BatchNormalizati (None, 4, 4, 256)    1024        res4b_branch2a[0][0]             __________________________________________________________________________________________________activation_26 (Activation)      (None, 4, 4, 256)    0           bn4b_branch2a[0][0]              __________________________________________________________________________________________________res4b_branch2b (Conv2D)         (None, 4, 4, 256)    590080      activation_26[0][0]              __________________________________________________________________________________________________bn4b_branch2b (BatchNormalizati (None, 4, 4, 256)    1024        res4b_branch2b[0][0]             __________________________________________________________________________________________________activation_27 (Activation)      (None, 4, 4, 256)    0           bn4b_branch2b[0][0]              __________________________________________________________________________________________________res4b_branch2c (Conv2D)         (None, 4, 4, 1024)   263168      activation_27[0][0]              __________________________________________________________________________________________________bn4b_branch2c (BatchNormalizati (None, 4, 4, 1024)   4096        res4b_branch2c[0][0]             __________________________________________________________________________________________________add_9 (Add)                     (None, 4, 4, 1024)   0           bn4b_branch2c[0][0]                                                                               activation_25[0][0]              __________________________________________________________________________________________________activation_28 (Activation)      (None, 4, 4, 1024)   0           add_9[0][0]                      __________________________________________________________________________________________________res4c_branch2a (Conv2D)         (None, 4, 4, 256)    262400      activation_28[0][0]              __________________________________________________________________________________________________bn4c_branch2a (BatchNormalizati (None, 4, 4, 256)    1024        res4c_branch2a[0][0]             __________________________________________________________________________________________________activation_29 (Activation)      (None, 4, 4, 256)    0           bn4c_branch2a[0][0]              __________________________________________________________________________________________________res4c_branch2b (Conv2D)         (None, 4, 4, 256)    590080      activation_29[0][0]              __________________________________________________________________________________________________bn4c_branch2b (BatchNormalizati (None, 4, 4, 256)    1024        res4c_branch2b[0][0]             __________________________________________________________________________________________________activation_30 (Activation)      (None, 4, 4, 256)    0           bn4c_branch2b[0][0]              __________________________________________________________________________________________________res4c_branch2c (Conv2D)         (None, 4, 4, 1024)   263168      activation_30[0][0]              __________________________________________________________________________________________________bn4c_branch2c (BatchNormalizati (None, 4, 4, 1024)   4096        res4c_branch2c[0][0]             __________________________________________________________________________________________________add_10 (Add)                    (None, 4, 4, 1024)   0           bn4c_branch2c[0][0]                                                                               activation_28[0][0]              __________________________________________________________________________________________________activation_31 (Activation)      (None, 4, 4, 1024)   0           add_10[0][0]                     __________________________________________________________________________________________________res4d_branch2a (Conv2D)         (None, 4, 4, 256)    262400      activation_31[0][0]              __________________________________________________________________________________________________bn4d_branch2a (BatchNormalizati (None, 4, 4, 256)    1024        res4d_branch2a[0][0]             __________________________________________________________________________________________________activation_32 (Activation)      (None, 4, 4, 256)    0           bn4d_branch2a[0][0]              __________________________________________________________________________________________________res4d_branch2b (Conv2D)         (None, 4, 4, 256)    590080      activation_32[0][0]              __________________________________________________________________________________________________bn4d_branch2b (BatchNormalizati (None, 4, 4, 256)    1024        res4d_branch2b[0][0]             __________________________________________________________________________________________________activation_33 (Activation)      (None, 4, 4, 256)    0           bn4d_branch2b[0][0]              __________________________________________________________________________________________________res4d_branch2c (Conv2D)         (None, 4, 4, 1024)   263168      activation_33[0][0]              __________________________________________________________________________________________________bn4d_branch2c (BatchNormalizati (None, 4, 4, 1024)   4096        res4d_branch2c[0][0]             __________________________________________________________________________________________________add_11 (Add)                    (None, 4, 4, 1024)   0           bn4d_branch2c[0][0]                                                                               activation_31[0][0]              __________________________________________________________________________________________________activation_34 (Activation)      (None, 4, 4, 1024)   0           add_11[0][0]                     __________________________________________________________________________________________________res4e_branch2a (Conv2D)         (None, 4, 4, 256)    262400      activation_34[0][0]              __________________________________________________________________________________________________bn4e_branch2a (BatchNormalizati (None, 4, 4, 256)    1024        res4e_branch2a[0][0]             __________________________________________________________________________________________________activation_35 (Activation)      (None, 4, 4, 256)    0           bn4e_branch2a[0][0]              __________________________________________________________________________________________________res4e_branch2b (Conv2D)         (None, 4, 4, 256)    590080      activation_35[0][0]              __________________________________________________________________________________________________bn4e_branch2b (BatchNormalizati (None, 4, 4, 256)    1024        res4e_branch2b[0][0]             __________________________________________________________________________________________________activation_36 (Activation)      (None, 4, 4, 256)    0           bn4e_branch2b[0][0]              __________________________________________________________________________________________________res4e_branch2c (Conv2D)         (None, 4, 4, 1024)   263168      activation_36[0][0]              __________________________________________________________________________________________________bn4e_branch2c (BatchNormalizati (None, 4, 4, 1024)   4096        res4e_branch2c[0][0]             __________________________________________________________________________________________________add_12 (Add)                    (None, 4, 4, 1024)   0           bn4e_branch2c[0][0]                                                                               activation_34[0][0]              __________________________________________________________________________________________________activation_37 (Activation)      (None, 4, 4, 1024)   0           add_12[0][0]                     __________________________________________________________________________________________________res4f_branch2a (Conv2D)         (None, 4, 4, 256)    262400      activation_37[0][0]              __________________________________________________________________________________________________bn4f_branch2a (BatchNormalizati (None, 4, 4, 256)    1024        res4f_branch2a[0][0]             __________________________________________________________________________________________________activation_38 (Activation)      (None, 4, 4, 256)    0           bn4f_branch2a[0][0]              __________________________________________________________________________________________________res4f_branch2b (Conv2D)         (None, 4, 4, 256)    590080      activation_38[0][0]              __________________________________________________________________________________________________bn4f_branch2b (BatchNormalizati (None, 4, 4, 256)    1024        res4f_branch2b[0][0]             __________________________________________________________________________________________________activation_39 (Activation)      (None, 4, 4, 256)    0           bn4f_branch2b[0][0]              __________________________________________________________________________________________________res4f_branch2c (Conv2D)         (None, 4, 4, 1024)   263168      activation_39[0][0]              __________________________________________________________________________________________________bn4f_branch2c (BatchNormalizati (None, 4, 4, 1024)   4096        res4f_branch2c[0][0]             __________________________________________________________________________________________________add_13 (Add)                    (None, 4, 4, 1024)   0           bn4f_branch2c[0][0]                                                                               activation_37[0][0]              __________________________________________________________________________________________________activation_40 (Activation)      (None, 4, 4, 1024)   0           add_13[0][0]                     __________________________________________________________________________________________________res5a_branch2a (Conv2D)         (None, 2, 2, 512)    524800      activation_40[0][0]              __________________________________________________________________________________________________bn5a_branch2a (BatchNormalizati (None, 2, 2, 512)    2048        res5a_branch2a[0][0]             __________________________________________________________________________________________________activation_41 (Activation)      (None, 2, 2, 512)    0           bn5a_branch2a[0][0]              __________________________________________________________________________________________________res5a_branch2b (Conv2D)         (None, 2, 2, 512)    2359808     activation_41[0][0]              __________________________________________________________________________________________________bn5a_branch2b (BatchNormalizati (None, 2, 2, 512)    2048        res5a_branch2b[0][0]             __________________________________________________________________________________________________activation_42 (Activation)      (None, 2, 2, 512)    0           bn5a_branch2b[0][0]              __________________________________________________________________________________________________res5a_branch2c (Conv2D)         (None, 2, 2, 2048)   1050624     activation_42[0][0]              __________________________________________________________________________________________________res5a_branch1 (Conv2D)          (None, 2, 2, 2048)   2099200     activation_40[0][0]              __________________________________________________________________________________________________bn5a_branch2c (BatchNormalizati (None, 2, 2, 2048)   8192        res5a_branch2c[0][0]             __________________________________________________________________________________________________bn5a_branch1 (BatchNormalizatio (None, 2, 2, 2048)   8192        res5a_branch1[0][0]              __________________________________________________________________________________________________add_14 (Add)                    (None, 2, 2, 2048)   0           bn5a_branch2c[0][0]                                                                               bn5a_branch1[0][0]               __________________________________________________________________________________________________activation_43 (Activation)      (None, 2, 2, 2048)   0           add_14[0][0]                     __________________________________________________________________________________________________res5b_branch2a (Conv2D)         (None, 2, 2, 512)    1049088     activation_43[0][0]              __________________________________________________________________________________________________bn5b_branch2a (BatchNormalizati (None, 2, 2, 512)    2048        res5b_branch2a[0][0]             __________________________________________________________________________________________________activation_44 (Activation)      (None, 2, 2, 512)    0           bn5b_branch2a[0][0]              __________________________________________________________________________________________________res5b_branch2b (Conv2D)         (None, 2, 2, 512)    2359808     activation_44[0][0]              __________________________________________________________________________________________________bn5b_branch2b (BatchNormalizati (None, 2, 2, 512)    2048        res5b_branch2b[0][0]             __________________________________________________________________________________________________activation_45 (Activation)      (None, 2, 2, 512)    0           bn5b_branch2b[0][0]              __________________________________________________________________________________________________res5b_branch2c (Conv2D)         (None, 2, 2, 2048)   1050624     activation_45[0][0]              __________________________________________________________________________________________________bn5b_branch2c (BatchNormalizati (None, 2, 2, 2048)   8192        res5b_branch2c[0][0]             __________________________________________________________________________________________________add_15 (Add)                    (None, 2, 2, 2048)   0           bn5b_branch2c[0][0]                                                                               activation_43[0][0]              __________________________________________________________________________________________________activation_46 (Activation)      (None, 2, 2, 2048)   0           add_15[0][0]                     __________________________________________________________________________________________________res5c_branch2a (Conv2D)         (None, 2, 2, 512)    1049088     activation_46[0][0]              __________________________________________________________________________________________________bn5c_branch2a (BatchNormalizati (None, 2, 2, 512)    2048        res5c_branch2a[0][0]             __________________________________________________________________________________________________activation_47 (Activation)      (None, 2, 2, 512)    0           bn5c_branch2a[0][0]              __________________________________________________________________________________________________res5c_branch2b (Conv2D)         (None, 2, 2, 512)    2359808     activation_47[0][0]              __________________________________________________________________________________________________bn5c_branch2b (BatchNormalizati (None, 2, 2, 512)    2048        res5c_branch2b[0][0]             __________________________________________________________________________________________________activation_48 (Activation)      (None, 2, 2, 512)    0           bn5c_branch2b[0][0]              __________________________________________________________________________________________________res5c_branch2c (Conv2D)         (None, 2, 2, 2048)   1050624     activation_48[0][0]              __________________________________________________________________________________________________bn5c_branch2c (BatchNormalizati (None, 2, 2, 2048)   8192        res5c_branch2c[0][0]             __________________________________________________________________________________________________add_16 (Add)                    (None, 2, 2, 2048)   0           bn5c_branch2c[0][0]                                                                               activation_46[0][0]              __________________________________________________________________________________________________activation_49 (Activation)      (None, 2, 2, 2048)   0           add_16[0][0]                     __________________________________________________________________________________________________average_pooling2d_1 (AveragePoo (None, 1, 1, 2048)   0           activation_49[0][0]              __________________________________________________________________________________________________flatten_1 (Flatten)             (None, 2048)         0           average_pooling2d_1[0][0]        __________________________________________________________________________________________________fc6 (Dense)                     (None, 6)            12294       flatten_1[0][0]                  ==================================================================================================Total params: 23,600,006Trainable params: 23,546,886Non-trainable params: 53,120__________________________________________________________________________________________________

深度学习之残差网络

最新关注

热文推荐

【计算讲谈社】第五讲 | 不止能上路，更能做好服务：自动驾驶产品规模化的问题定义

微软官宣放出一个「小模型」，仅2.7B参数，击败Llama2和Gemini Nano 2

UML设计系列(8)：数据库关联关系图

【陈老板赠书活动 – 10期】- 【Python之光：Python编程入门与实战】

【SpringMVC】| RESTful架构风格、RESTful案例(CRUD)

研究生复试简历模板及考研·面试·范文·经验

深度学习之残差网络

相关文章

最新关注

热文推荐