吴恩达深度学习课程deeplearning.ai课程作业：Class 4 Week 2 Residual Networks

吴恩达deeplearning.ai课程作业，自己写的答案。
由于这部分的作业后面要自己训练一个ResNet-50的网络，训练耗时较长。如果是CPU模式，一个epoch大约100s，但在我的GPU服务器上，一次epoch大约5s。不使用GPU训练，实在是耗时太长，没有GPU的话建议还是先用训练好的模型。我训练好的几个模型，下面给出百度云链接。
resnet50_20_epochs.h5 链接：https://pan.baidu.com/s/1eROf3BO 密码：qed2
resnet50_30_epochs.h5 链接：https://pan.baidu.com/s/1o8kPNUM 密码：tqio
resnet50_44_epochs.h5 链接：https://pan.baidu.com/s/1c1N3AzI 密码：2xwu
resnet50_55_epochs.h5 链接：https://pan.baidu.com/s/1bpfMA0v 密码：cxcv
Coursera上提供的模型文件：
ResNet50.h5 链接：链接：https://pan.baidu.com/s/1boCG2Iz 密码：sefq

Residual Networks

Welcome to the second assignment of this week! You will learn how to build very deep convolutional networks, using Residual Networks (ResNets). In theory, very deep networks can represent very complex functions; but in practice, they are hard to train. Residual Networks, introduced by He et al., allow you to train much deeper networks than were previously practically feasible.

In this assignment, you will:
- Implement the basic building blocks of ResNets.
- Put together these building blocks to implement and train a state-of-the-art neural network for image classification.

This assignment will be done in Keras.

Before jumping into the problem, let’s run the cell below to load the required packages.

import numpy as npfrom keras import layersfrom keras.layers import Input, Add, Dense, Activation, ZeroPadding2D, BatchNormalization, Flatten, Conv2D, AveragePooling2D, MaxPooling2D, GlobalMaxPooling2Dfrom keras.models import Model, load_modelfrom keras.preprocessing import imagefrom keras.utils import layer_utilsfrom keras.utils.data_utils import get_filefrom keras.applications.imagenet_utils import preprocess_inputimport pydotfrom IPython.display import SVGfrom keras.utils.vis_utils import model_to_dotfrom keras.utils import plot_modelfrom resnets_utils import *from keras.initializers import glorot_uniformimport scipy.miscfrom matplotlib.pyplot import imshow%matplotlib inlineimport keras.backend as KK.set_image_data_format('channels_last')K.set_learning_phase(1)

Using TensorFlow backend.

1 - The problem of very deep neural networks

Last week, you built your first convolutional neural network. In recent years, neural networks have become deeper, with state-of-the-art networks going from just a few layers (e.g., AlexNet) to over a hundred layers.

The main benefit of a very deep network is that it can represent very complex functions. It can also learn features at many different levels of abstraction, from edges (at the lower layers) to very complex features (at the deeper layers). However, using a deeper network doesn’t always help. A huge barrier to training them is vanishing gradients: very deep networks often have a gradient signal that goes to zero quickly, thus making gradient descent unbearably slow. More specifically, during gradient descent, as you backprop from the final layer back to the first layer, you are multiplying by the weight matrix on each step, and thus the gradient can decrease exponentially quickly to zero (or, in rare cases, grow exponentially quickly and “explode” to take very large values).

During training, you might therefore see the magnitude (or norm) of the gradient for the earlier layers descrease to zero very rapidly as training proceeds:

Figure 1 : Vanishing gradient
The speed of learning decreases very rapidly for the early layers as the network trains

You are now going to solve this problem by building a Residual Network!

2 - Building a Residual Network

In ResNets, a “shortcut” or a “skip connection” allows the gradient to be directly backpropagated to earlier layers:

Figure 2 : A ResNet block showing a skip-connection

The image on the left shows the “main path” through the network. The image on the right adds a shortcut to the main path. By stacking these ResNet blocks on top of each other, you can form a very deep network.

We also saw in lecture that having ResNet blocks with the shortcut also makes it very easy for one of the blocks to learn an identity function. This means that you can stack on additional ResNet blocks with little risk of harming training set performance. (There is also some evidence that the ease of learning an identity function–even more than skip connections helping with vanishing gradients–accounts for ResNets’ remarkable performance.)

Two main types of blocks are used in a ResNet, depending mainly on whether the input/output dimensions are same or different. You are going to implement both of them.

2.1 - The identity block

The identity block is the standard block used in ResNets, and corresponds to the case where the input activation (say a[l]) has the same dimension as the output activation (say a[l+2]). To flesh out the different steps of what happens in a ResNet’s identity block, here is an alternative diagram showing the individual steps:

Figure 3 : Identity block. Skip connection “skips over” 2 layers.

The upper path is the “shortcut path.” The lower path is the “main path.” In this diagram, we have also made explicit the CONV2D and ReLU steps in each layer. To speed up training we have also added a BatchNorm step. Don’t worry about this being complicated to implement–you’ll see that BatchNorm is just one line of code in Keras!

In this exercise, you’ll actually implement a slightly more powerful version of this identity block, in which the skip connection “skips over” 3 hidden layers rather than 2 layers. It looks like this:

Figure 4 : Identity block. Skip connection “skips over” 3 layers.

Here’re the individual steps.

First component of main path:
- The first CONV2D has F1 filters of shape (1,1) and a stride of (1,1). Its padding is “valid” and its name should be conv_name_base + '2a'. Use 0 as the seed for the random initialization.
- The first BatchNorm is normalizing the channels axis. Its name should be bn_name_base + '2a'.
- Then apply the ReLU activation function. This has no name and no hyperparameters.

Second component of main path:
- The second CONV2D has F2 filters of shape (f,f) and a stride of (1,1). Its padding is “same” and its name should be conv_name_base + '2b'. Use 0 as the seed for the random initialization.
- The second BatchNorm is normalizing the channels axis. Its name should be bn_name_base + '2b'.
- Then apply the ReLU activation function. This has no name and no hyperparameters.

Third component of main path:
- The third CONV2D has F3 filters of shape (1,1) and a stride of (1,1). Its padding is “valid” and its name should be conv_name_base + '2c'. Use 0 as the seed for the random initialization.
- The third BatchNorm is normalizing the channels axis. Its name should be bn_name_base + '2c'. Note that there is no ReLU activation function in this component.

Final step:
- The shortcut and the input are added together.
- Then apply the ReLU activation function. This has no name and no hyperparameters.

Exercise: Implement the ResNet identity block. We have implemented the first component of the main path. Please read over this carefully to make sure you understand what it is doing. You should implement the rest.
- To implement the Conv2D step: See reference
- To implement BatchNorm: See reference (axis: Integer, the axis that should be normalized (typically the channels axis))
- For the activation, use: Activation('relu')(X)
- To add the value passed forward by the shortcut: See reference

# GRADED FUNCTION: identity_blockdef identity_block(X, f, filters, stage, block):    """    Implementation of the identity block as defined in Figure 3    Arguments:    X -- input tensor of shape (m, n_H_prev, n_W_prev, n_C_prev)    f -- integer, specifying the shape of the middle CONV's window for the main path    filters -- python list of integers, defining the number of filters in the CONV layers of the main path    stage -- integer, used to name the layers, depending on their position in the network    block -- string/character, used to name the layers, depending on their position in the network    Returns:    X -- output of the identity block, tensor of shape (n_H, n_W, n_C)    """    # defining name basis    conv_name_base = 'res' + str(stage) + block + '_branch'    bn_name_base = 'bn' + str(stage) + block + '_branch'    # 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    # First component of main path    X = Conv2D(filters = F1, kernel_size = (1, 1), strides = (1,1), padding = 'valid', name = conv_name_base + '2a', kernel_initializer = glorot_uniform(seed=0))(X)    X = BatchNormalization(axis = 3, name = bn_name_base + '2a')(X)    X = Activation('relu')(X)    ### START CODE HERE ###    # Second component of main path (≈3 lines)    X = Conv2D(filters=F2, kernel_size=(f,f), strides=(1,1), padding='same', name=conv_name_base+'2b', kernel_initializer=glorot_uniform(seed=0))(X)    X = BatchNormalization(axis=3, name=bn_name_base + '2b')(X)    X = Activation('relu')(X)    # Third component of main path (≈2 lines)    X = Conv2D(filters=F3, kernel_size=(1,1), strides=(1,1), padding='valid', name=conv_name_base + '2c', kernel_initializer=glorot_uniform(seed=0))(X)    X = BatchNormalization(axis=3, name=bn_name_base + '2c')(X)    # Final step: Add shortcut value to main path, and pass it through a RELU activation (≈2 lines)    X = Add()([X, X_shortcut])    X = Activation('relu')(X)    ### END CODE HERE ###    return X

tf.reset_default_graph()with tf.Session() as test:    np.random.seed(1)    A_prev = tf.placeholder("float", [3, 4, 4, 6])    X = np.random.randn(3, 4, 4, 6)    A = identity_block(A_prev, f = 2, filters = [2, 4, 6], stage = 1, block = 'a')    test.run(tf.global_variables_initializer())    out = test.run([A], feed_dict={A_prev: X, K.learning_phase(): 0})    print("out = " + str(out[0][1][1][0]))

out = [ 0.94822997  0.          1.16101444  2.747859    0.          1.36677003]

Expected Output:

out [ 0.94822985 0. 1.16101444 2.747859 0. 1.36677003]

2.2 - The convolutional block

You’ve implemented the ResNet identity block. Next, the ResNet “convolutional block” is the other type of block. You can use this type of block when the input and output dimensions don’t match up. The difference with the identity block is that there is a CONV2D layer in the shortcut path:

Figure 4 : Convolutional block

The CONV2D layer in the shortcut path is used to resize the input x to a different dimension, so that the dimensions match up in the final addition needed to add the shortcut value back to the main path. (This plays a similar role as the matrix Ws discussed in lecture.) For example, to reduce the activation dimensions’s height and width by a factor of 2, you can use a 1x1 convolution with a stride of 2. The CONV2D layer on the shortcut path does not use any non-linear activation function. Its main role is to just apply a (learned) linear function that reduces the dimension of the input, so that the dimensions match up for the later addition step.

The details of the convolutional block are as follows.

First component of main path:
- The first CONV2D has F1 filters of shape (1,1) and a stride of (s,s). Its padding is “valid” and its name should be conv_name_base + '2a'.
- The first BatchNorm is normalizing the channels axis. Its name should be bn_name_base + '2a'.
- Then apply the ReLU activation function. This has no name and no hyperparameters.

Second component of main path:
- The second CONV2D has F2 filters of (f,f) and a stride of (1,1). Its padding is “same” and it’s name should be conv_name_base + '2b'.
- The second BatchNorm is normalizing the channels axis. Its name should be bn_name_base + '2b'.
- Then apply the ReLU activation function. This has no name and no hyperparameters.

Third component of main path:
- The third CONV2D has F3 filters of (1,1) and a stride of (1,1). Its padding is “valid” and it’s name should be conv_name_base + '2c'.
- The third BatchNorm is normalizing the channels axis. Its name should be bn_name_base + '2c'. Note that there is no ReLU activation function in this component.

Shortcut path:
- The CONV2D has F3 filters of shape (1,1) and a stride of (s,s). Its padding is “valid” and its name should be conv_name_base + '1'.
- The BatchNorm is normalizing the channels axis. Its name should be bn_name_base + '1'.

Final step:
- The shortcut and the main path values are added together.
- Then apply the ReLU activation function. This has no name and no hyperparameters.

Exercise: Implement the convolutional block. We have implemented the first component of the main path; you should implement the rest. As before, always use 0 as the seed for the random initialization, to ensure consistency with our grader.
- Conv Hint
- BatchNorm Hint (axis: Integer, the axis that should be normalized (typically the features axis))
- For the activation, use: Activation('relu')(X)
- Addition Hint

# GRADED FUNCTION: convolutional_blockdef convolutional_block(X, f, filters, stage, block, s = 2):    """    Implementation of the convolutional block as defined in Figure 4    Arguments:    X -- input tensor of shape (m, n_H_prev, n_W_prev, n_C_prev)    f -- integer, specifying the shape of the middle CONV's window for the main path    filters -- python list of integers, defining the number of filters in the CONV layers of the main path    stage -- integer, used to name the layers, depending on their position in the network    block -- string/character, used to name the layers, depending on their position in the network    s -- Integer, specifying the stride to be used    Returns:    X -- output of the convolutional block, tensor of shape (n_H, n_W, n_C)    """    # defining name basis    conv_name_base = 'res' + str(stage) + block + '_branch'    bn_name_base = 'bn' + str(stage) + block + '_branch'    # Retrieve Filters    F1, F2, F3 = filters    # Save the input value    X_shortcut = X    ##### MAIN PATH #####    # First component of main path     X = Conv2D(F1, (1, 1), strides = (s,s), name = conv_name_base + '2a', kernel_initializer = glorot_uniform(seed=0))(X)    X = BatchNormalization(axis = 3, name = bn_name_base + '2a')(X)    X = Activation('relu')(X)    ### START CODE HERE ###    # Second component of main path (≈3 lines)    X = Conv2D(filters=F2, kernel_size=(f,f), strides=(1,1), padding='same', name=conv_name_base+'2b', kernel_initializer=glorot_uniform(seed=0))(X)    X = BatchNormalization(axis=3, name=bn_name_base+'2b')(X)    X = Activation('relu')(X)    # Third component of main path (≈2 lines)    X = Conv2D(filters=F3, kernel_size=(1,1), strides=(1,1), padding='valid', name=conv_name_base+'2c', kernel_initializer=glorot_uniform(seed=0))(X)    X = BatchNormalization(axis=3, name=bn_name_base+'2c')(X)    ##### SHORTCUT PATH #### (≈2 lines)    X_shortcut = Conv2D(filters=F3, kernel_size=(1,1), strides=(s, s), padding='valid', name=conv_name_base+'1', kernel_initializer=glorot_uniform(seed=0))(X_shortcut)    X_shortcut = BatchNormalization(axis=3, name=bn_name_base+'1')(X_shortcut)    # Final step: Add shortcut value to main path, and pass it through a RELU activation (≈2 lines)    X = Add()([X, X_shortcut])    X = Activation('relu')(X)    ### END CODE HERE ###    return X

tf.reset_default_graph()with tf.Session() as test:    np.random.seed(1)    A_prev = tf.placeholder("float", [3, 4, 4, 6])    X = np.random.randn(3, 4, 4, 6)    A = convolutional_block(A_prev, f = 2, filters = [2, 4, 6], stage = 1, block = 'a')    test.run(tf.global_variables_initializer())    out = test.run([A], feed_dict={A_prev: X, K.learning_phase(): 0})#     print(len(out[0]))#     print(out)    print("out = " + str(out[0][1][1][0]))

out = [ 0.09018461  1.23489773  0.46822017  0.0367176   0.          0.65516603]

Expected Output:

out [ 0.09018463 1.23489773 0.46822017 0.0367176 0. 0.65516603]

3 - Building your first ResNet model (50 layers)

You now have the necessary blocks to build a very deep ResNet. The following figure describes in detail the architecture of this neural network. “ID BLOCK” in the diagram stands for “Identity block,” and “ID BLOCK x3” means you should stack 3 identity blocks together.

Figure 5 : ResNet-50 model

The details of this ResNet-50 model are:
- Zero-padding pads the input with a pad of (3,3)
- Stage 1:
- The 2D Convolution has 64 filters of shape (7,7) and uses a stride of (2,2). Its name is “conv1”.
- BatchNorm is applied to the channels axis of the input.
- MaxPooling uses a (3,3) window and a (2,2) stride.
- Stage 2:
- The convolutional block uses three set of filters of size [64,64,256], “f” is 3, “s” is 1 and the block is “a”.
- The 2 identity blocks use three set of filters of size [64,64,256], “f” is 3 and the blocks are “b” and “c”.
- Stage 3:
- The convolutional block uses three set of filters of size [128,128,512], “f” is 3, “s” is 2 and the block is “a”.
- The 3 identity blocks use three set of filters of size [128,128,512], “f” is 3 and the blocks are “b”, “c” and “d”.
- Stage 4:
- The convolutional block uses three set of filters of size [256, 256, 1024], “f” is 3, “s” is 2 and the block is “a”.
- The 5 identity blocks use three set of filters of size [256, 256, 1024], “f” is 3 and the blocks are “b”, “c”, “d”, “e” and “f”.
- Stage 5:
- The convolutional block uses three set of filters of size [512, 512, 2048], “f” is 3, “s” is 2 and the block is “a”.
- The 2 identity blocks use three set of filters of size [256, 256, 2048], “f” is 3 and the blocks are “b” and “c”.
- The 2D Average Pooling uses a window of shape (2,2) and its name is “avg_pool”.
- The flatten doesn’t have any hyperparameters or name.
- The Fully Connected (Dense) layer reduces its input to the number of classes using a softmax activation. Its name should be 'fc' + str(classes).

Exercise: Implement the ResNet with 50 layers described in the figure above. We have implemented Stages 1 and 2. Please implement the rest. (The syntax for implementing Stages 3-5 should be quite similar to that of Stage 2.) Make sure you follow the naming convention in the text above.

You’ll need to use this function:
- Average pooling see reference

Here’re some other functions we used in the code below:
- Conv2D: See reference
- BatchNorm: See reference (axis: Integer, the axis that should be normalized (typically the features axis))
- Zero padding: See reference
- Max pooling: See reference
- Fully conected layer: See reference
- Addition: See reference

# GRADED FUNCTION: ResNet50def ResNet50(input_shape = (64, 64, 3), classes = 6):    """    Implementation of the popular ResNet50 the following architecture:    CONV2D -> BATCHNORM -> RELU -> MAXPOOL -> CONVBLOCK -> IDBLOCK*2 -> CONVBLOCK -> IDBLOCK*3    -> CONVBLOCK -> IDBLOCK*5 -> CONVBLOCK -> IDBLOCK*2 -> AVGPOOL -> TOPLAYER    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), name = 'conv1', kernel_initializer = glorot_uniform(seed=0))(X)    X = BatchNormalization(axis = 3, name = 'bn_conv1')(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], stage = 2, block='a', s = 1)    X = identity_block(X, 3, [64, 64, 256], stage=2, block='b')    X = identity_block(X, 3, [64, 64, 256], stage=2, block='c')    ### START CODE HERE ###    # helper functions    # convolutional_block(X, f, filters, stage, block, s = 2)    # identity_block(X, f, filters, stage, block)    # Stage 3 (≈4 lines)    X = convolutional_block(X, f=3, filters=[128, 128, 512], stage=3, block='a', s=2)    X = identity_block(X, f=3, filters=[128, 128, 512], stage=3, block='b')    X = identity_block(X, f=3, filters=[128, 128, 512], stage=3, block='c')    X = identity_block(X, f=3, filters=[128, 128, 512], stage=3, block='d')    # Stage 4 (≈6 lines)    X = convolutional_block(X, f=3, filters=[256, 256, 1024], stage=4, block='a', s=2)    X = identity_block(X, f=3, filters=[256, 256, 1024], stage=4, block='b')    X = identity_block(X, f=3, filters=[256, 256, 1024], stage=4, block='c')    X = identity_block(X, f=3, filters=[256, 256, 1024], stage=4, block='d')    X = identity_block(X, f=3, filters=[256, 256, 1024], stage=4, block='e')    X = identity_block(X, f=3, filters=[256, 256, 1024], stage=4, block='f')    # Stage 5 (≈3 lines)    X = convolutional_block(X, f=3, filters=[512, 512, 2048], stage=5, block='a', s=2)    X = identity_block(X, f=3, filters=[512, 512, 2048], stage=5, block='b')    X = identity_block(X, f=3, filters=[512, 512, 2048], stage=5, block='c')    # AVGPOOL (≈1 line). Use "X = AveragePooling2D(...)(X)"    X = AveragePooling2D((2,2), name='avg_pool')(X)    ### END CODE HERE ###    # output layer    X = Flatten()(X)    X = Dense(classes, activation='softmax', name='fc' + str(classes), kernel_initializer = glorot_uniform(seed=0))(X)    # Create model    model = Model(inputs = X_input, outputs = X, name='ResNet50')    return model

Run the following code to build the model’s graph. If your implementation is not correct you will know it by checking your accuracy when running model.fit(...) below.

model = ResNet50(input_shape = (64, 64, 3), classes = 6)

As seen in the Keras Tutorial Notebook, prior training a model, you need to configure the learning process by compiling the model.

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

The model is now ready to be trained. The only thing you need is a dataset.

Let’s load the SIGNS Dataset.

Figure 6 : SIGNS dataset

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)

Run the following cell to train your model on 2 epochs with a batch size of 32. On a CPU it should take you around 5min per epoch.

model.fit(X_train, Y_train, epochs = 2, batch_size = 32)

Epoch 1/21080/1080 [==============================] - 107s 99ms/step - loss: 3.0556 - acc: 0.2481Epoch 2/21080/1080 [==============================] - 103s 95ms/step - loss: 2.4399 - acc: 0.3278<keras.callbacks.History at 0x7ff1c149af60>

Expected Output:

Epoch 1/2 loss: between 1 and 5, acc: between 0.2 and 0.5, although your results can be different from ours. Epoch 2/2 loss: between 1 and 5, acc: between 0.2 and 0.5, you should see your loss decreasing and the accuracy increasing.

Let’s see how this model (trained on only two epochs) performs on the test set.

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

120/120 [==============================] - 5s 38ms/stepLoss = 2.24901695251Test Accuracy = 0.166666666667

Expected Output:

Test Accuracy between 0.16 and 0.25

For the purpose of this assignment, we’ve asked you to train the model only for two epochs. You can see that it achieves poor performances. Please go ahead and submit your assignment; to check correctness, the online grader will run your code only for a small number of epochs as well.

After you have finished this official (graded) part of this assignment, you can also optionally train the ResNet for more iterations, if you want. We get a lot better performance when we train for ~20 epochs, but this will take more than an hour when training on a CPU.

Using a GPU, we’ve trained our own ResNet50 model’s weights on the SIGNS dataset. You can load and run our trained model on the test set in the cells below. It may take ≈1min to load the model.

注：这里我导入的是自己在GPU服务器上训练的模型，故修改成了对应的名字。

# model = load_model('ResNet50.h5') model = load_model('resnet50_44_epochs.h5') 

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

120/120 [==============================] - 9s 78ms/stepLoss = 0.0914498666922Test Accuracy = 0.958333337307

ResNet50 is a powerful model for image classification when it is trained for an adequate number of iterations. We hope you can use what you’ve learnt and apply it to your own classification problem to perform state-of-the-art accuracy.

Congratulations on finishing this assignment! You’ve now implemented a state-of-the-art image classification system!

4 - Test on your own image (Optional/Ungraded)

If you wish, you can also take a picture of your own hand and see the output of the model. To do this:
1. Click on “File” in the upper bar of this notebook, then click “Open” to go on your Coursera Hub.
2. Add your image to this Jupyter Notebook’s directory, in the “images” folder
3. Write your image’s name in the following code
4. Run the code and check if the algorithm is right!

img_path = 'images/my_image.jpg'img = image.load_img(img_path, target_size=(64, 64))x = image.img_to_array(img)x = np.expand_dims(x, axis=0)x = preprocess_input(x)print('Input image shape:', x.shape)my_image = scipy.misc.imread(img_path)imshow(my_image)print("class prediction vector [p(0), p(1), p(2), p(3), p(4), p(5)] = ")print(model.predict(x))

Input image shape: (1, 64, 64, 3)class prediction vector [p(0), p(1), p(2), p(3), p(4), p(5)] = [[ 1.  0.  0.  0.  0.  0.]]

You can also print a summary of your model by running the following code.

model.summary()

（打印出resnet的网络结构，省略）

Finally, run the code below to visualize your ResNet50. You can also download a .png picture of your model by going to “File -> Open…-> model.png”.

plot_model(model, to_file='model.png')SVG(model_to_dot(model).create(prog='dot', format='svg'))

（画出图表表示resnet，省略）

What you should remember:
- Very deep “plain” networks don’t work in practice because they are hard to train due to vanishing gradients.
- The skip-connections help to address the Vanishing Gradient problem. They also make it easy for a ResNet block to learn an identity function.
- There are two main type of blocks: The identity block and the convolutional block.
- Very deep Residual Networks are built by stacking these blocks together.

References

This notebook presents the ResNet algorithm due to He et al. (2015). The implementation here also took significant inspiration and follows the structure given in the github repository of Francois Chollet:

• Kaiming He, Xiangyu Zhang, Shaoqing Ren, Jian Sun - Deep Residual Learning for Image Recognition (2015)
• Francois Chollet’s github repository: https://github.com/fchollet/deep-learning-models/blob/master/resnet50.py

训练过程中的代码

epoch = 20

model1 = ResNet50(input_shape = (64, 64, 3), classes = 6)model1.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])model1.fit(X_train, Y_train, epochs = 20, batch_size = 32)model1.save('resnet50_20_epochs.h5')preds = model1.evaluate(X_test, Y_test)print ("Loss = " + str(preds[0]))print ("Test Accuracy = " + str(preds[1]))

Epoch 1/201080/1080 [==============================] - 15s 14ms/step - loss: 2.5141 - acc: 0.4241Epoch 2/201080/1080 [==============================] - 5s 5ms/step - loss: 1.7727 - acc: 0.6194Epoch 3/201080/1080 [==============================] - 6s 5ms/step - loss: 1.4935 - acc: 0.6769Epoch 4/201080/1080 [==============================] - 5s 5ms/step - loss: 1.5494 - acc: 0.5833Epoch 5/201080/1080 [==============================] - 5s 5ms/step - loss: 0.6902 - acc: 0.7889Epoch 6/201080/1080 [==============================] - 5s 5ms/step - loss: 0.4155 - acc: 0.8593Epoch 7/201080/1080 [==============================] - 5s 5ms/step - loss: 0.2782 - acc: 0.9139Epoch 8/201080/1080 [==============================] - 5s 5ms/step - loss: 0.1665 - acc: 0.9500Epoch 9/201080/1080 [==============================] - 5s 5ms/step - loss: 0.2578 - acc: 0.9185Epoch 10/201080/1080 [==============================] - 5s 5ms/step - loss: 0.1690 - acc: 0.9435Epoch 11/201080/1080 [==============================] - 5s 5ms/step - loss: 0.0913 - acc: 0.9694Epoch 12/201080/1080 [==============================] - 5s 5ms/step - loss: 0.1389 - acc: 0.9602Epoch 13/201080/1080 [==============================] - 5s 5ms/step - loss: 0.1490 - acc: 0.9444Epoch 14/201080/1080 [==============================] - 5s 5ms/step - loss: 0.1044 - acc: 0.9694Epoch 15/201080/1080 [==============================] - 5s 5ms/step - loss: 0.0435 - acc: 0.9861Epoch 16/201080/1080 [==============================] - 5s 5ms/step - loss: 0.0324 - acc: 0.9926Epoch 17/201080/1080 [==============================] - 5s 5ms/step - loss: 0.0190 - acc: 0.9926Epoch 18/201080/1080 [==============================] - 5s 5ms/step - loss: 0.0577 - acc: 0.9824Epoch 19/201080/1080 [==============================] - 5s 5ms/step - loss: 0.0268 - acc: 0.9907Epoch 20/201080/1080 [==============================] - 5s 5ms/step - loss: 0.0662 - acc: 0.9787120/120 [==============================] - 2s 17ms/stepLoss = 0.825686124961Test Accuracy = 0.833333333333

epoch = 50

model2 = ResNet50(input_shape = (64, 64, 3), classes = 6)model2.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])model2.fit(X_train, Y_train, epochs = 50, batch_size = 32)model2.save('resnet50_50_epochs.h5')preds = model2.evaluate(X_test, Y_test)print ("Loss = " + str(preds[0]))print ("Test Accuracy = " + str(preds[1]))

Epoch 1/501080/1080 [==============================] - 18s 17ms/step - loss: 2.1776 - acc: 0.4556Epoch 2/501080/1080 [==============================] - 5s 5ms/step - loss: 1.8498 - acc: 0.5370Epoch 3/501080/1080 [==============================] - 5s 5ms/step - loss: 0.9010 - acc: 0.6852Epoch 4/501080/1080 [==============================] - 5s 5ms/step - loss: 0.4735 - acc: 0.8407Epoch 5/501080/1080 [==============================] - 5s 5ms/step - loss: 0.2245 - acc: 0.9222Epoch 6/501080/1080 [==============================] - 5s 5ms/step - loss: 0.1450 - acc: 0.9611Epoch 7/501080/1080 [==============================] - 5s 5ms/step - loss: 0.7002 - acc: 0.7759Epoch 8/501080/1080 [==============================] - 5s 5ms/step - loss: 0.2651 - acc: 0.9102Epoch 9/501080/1080 [==============================] - 5s 5ms/step - loss: 0.1757 - acc: 0.9481Epoch 10/501080/1080 [==============================] - 5s 5ms/step - loss: 0.1131 - acc: 0.9602Epoch 11/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0816 - acc: 0.9759Epoch 12/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0332 - acc: 0.9907Epoch 13/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0397 - acc: 0.9861Epoch 14/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0305 - acc: 0.9907Epoch 15/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0318 - acc: 0.9889Epoch 16/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0125 - acc: 0.9972Epoch 17/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0279 - acc: 0.9907Epoch 18/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0888 - acc: 0.9657Epoch 19/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0460 - acc: 0.9843Epoch 20/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0512 - acc: 0.9787Epoch 21/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0423 - acc: 0.9843Epoch 22/501080/1080 [==============================] - 6s 5ms/step - loss: 0.0473 - acc: 0.9870Epoch 23/501080/1080 [==============================] - 5s 5ms/step - loss: 0.1245 - acc: 0.9750Epoch 24/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0739 - acc: 0.9741Epoch 25/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0663 - acc: 0.9815Epoch 26/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0175 - acc: 0.9926Epoch 27/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0103 - acc: 0.9981Epoch 28/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0496 - acc: 0.9963Epoch 29/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0023 - acc: 0.9991Epoch 30/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0085 - acc: 0.9972Epoch 31/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0329 - acc: 0.9981Epoch 32/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0039 - acc: 0.9981Epoch 33/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0071 - acc: 0.9981Epoch 34/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0237 - acc: 0.9898Epoch 35/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0667 - acc: 0.9769Epoch 36/501080/1080 [==============================] - 5s 5ms/step - loss: 0.1863 - acc: 0.9500Epoch 37/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0612 - acc: 0.9787Epoch 38/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0362 - acc: 0.9880Epoch 39/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0251 - acc: 0.9935Epoch 40/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0210 - acc: 0.9898Epoch 41/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0090 - acc: 0.9981Epoch 42/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0534 - acc: 0.9870Epoch 43/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0138 - acc: 0.9954Epoch 44/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0055 - acc: 0.9991Epoch 45/501080/1080 [==============================] - 5s 5ms/step - loss: 4.9548e-04 - acc: 1.0000Epoch 46/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0015 - acc: 1.0000Epoch 47/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0086 - acc: 0.9972Epoch 48/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0213 - acc: 0.9944Epoch 49/501080/1080 [==============================] - 5s 5ms/step - loss: 0.0286 - acc: 0.9898Epoch 50/501080/1080 [==============================] - 6s 5ms/step - loss: 0.0228 - acc: 0.9880120/120 [==============================] - 4s 31ms/stepLoss = 1.81704656283Test Accuracy = 0.69166667064

epoch = 30

model3 = ResNet50(input_shape = (64, 64, 3), classes = 6)model3.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])model3.fit(X_train, Y_train, epochs = 30, batch_size = 32)model3.save('resnet50_30_epochs.h5')preds = model3.evaluate(X_test, Y_test)print ("Loss = " + str(preds[0]))print ("Test Accuracy = " + str(preds[1]))

Epoch 1/301080/1080 [==============================] - 13s 12ms/step - loss: 1.9815 - acc: 0.4713Epoch 2/301080/1080 [==============================] - 5s 5ms/step - loss: 0.5419 - acc: 0.8120Epoch 3/301080/1080 [==============================] - 5s 5ms/step - loss: 0.4135 - acc: 0.8713Epoch 4/301080/1080 [==============================] - 5s 5ms/step - loss: 0.4284 - acc: 0.8713Epoch 5/301080/1080 [==============================] - 5s 5ms/step - loss: 0.4162 - acc: 0.8722Epoch 6/301080/1080 [==============================] - 5s 5ms/step - loss: 0.1329 - acc: 0.9546Epoch 7/301080/1080 [==============================] - 5s 5ms/step - loss: 0.1342 - acc: 0.9602Epoch 8/301080/1080 [==============================] - 5s 5ms/step - loss: 0.1336 - acc: 0.9630Epoch 9/301080/1080 [==============================] - 5s 5ms/step - loss: 0.1799 - acc: 0.9611Epoch 10/301080/1080 [==============================] - 5s 5ms/step - loss: 0.4936 - acc: 0.8731Epoch 11/301080/1080 [==============================] - 5s 5ms/step - loss: 0.5063 - acc: 0.8398Epoch 12/301080/1080 [==============================] - 5s 5ms/step - loss: 0.1922 - acc: 0.9426Epoch 13/301080/1080 [==============================] - 5s 5ms/step - loss: 0.2066 - acc: 0.9435Epoch 14/301080/1080 [==============================] - 5s 5ms/step - loss: 0.2782 - acc: 0.9139Epoch 15/301080/1080 [==============================] - 5s 5ms/step - loss: 0.2053 - acc: 0.9324Epoch 16/301080/1080 [==============================] - 5s 5ms/step - loss: 0.1632 - acc: 0.9602Epoch 17/301080/1080 [==============================] - 5s 5ms/step - loss: 0.0824 - acc: 0.9787Epoch 18/301080/1080 [==============================] - 5s 5ms/step - loss: 0.0485 - acc: 0.9880Epoch 19/301080/1080 [==============================] - 5s 5ms/step - loss: 0.0145 - acc: 0.9981Epoch 20/301080/1080 [==============================] - 5s 5ms/step - loss: 0.0145 - acc: 0.9944Epoch 21/301080/1080 [==============================] - 5s 5ms/step - loss: 0.0119 - acc: 0.9963Epoch 22/301080/1080 [==============================] - 5s 5ms/step - loss: 0.0280 - acc: 0.9954Epoch 23/301080/1080 [==============================] - 5s 5ms/step - loss: 0.0187 - acc: 0.9917Epoch 24/301080/1080 [==============================] - 5s 5ms/step - loss: 0.0483 - acc: 0.9824Epoch 25/301080/1080 [==============================] - 5s 5ms/step - loss: 0.0939 - acc: 0.9685Epoch 26/301080/1080 [==============================] - 5s 5ms/step - loss: 0.0390 - acc: 0.9907Epoch 27/301080/1080 [==============================] - 5s 5ms/step - loss: 0.0197 - acc: 0.9917Epoch 28/301080/1080 [==============================] - 5s 5ms/step - loss: 0.0270 - acc: 0.9972Epoch 29/301080/1080 [==============================] - 5s 5ms/step - loss: 0.0021 - acc: 1.0000Epoch 30/301080/1080 [==============================] - 5s 5ms/step - loss: 0.0012 - acc: 1.0000120/120 [==============================] - 1s 11ms/stepLoss = 0.0633144895857Test Accuracy = 0.983333333333

epoch = 44

model4 = ResNet50(input_shape = (64, 64, 3), classes = 6)model4.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])model4.fit(X_train, Y_train, epochs = 44, batch_size = 32)model4.save('resnet50_44_epochs.h5')preds = model3.evaluate(X_test, Y_test)print ("Loss = " + str(preds[0]))print ("Test Accuracy = " + str(preds[1]))

Epoch 1/441080/1080 [==============================] - 30s 28ms/step - loss: 2.1430 - acc: 0.3972Epoch 2/441080/1080 [==============================] - 5s 5ms/step - loss: 0.9240 - acc: 0.6972Epoch 3/441080/1080 [==============================] - 5s 5ms/step - loss: 0.3779 - acc: 0.8583Epoch 4/441080/1080 [==============================] - 5s 5ms/step - loss: 0.2152 - acc: 0.9352Epoch 5/441080/1080 [==============================] - 5s 5ms/step - loss: 0.1680 - acc: 0.9463Epoch 6/441080/1080 [==============================] - 5s 5ms/step - loss: 0.3060 - acc: 0.9065Epoch 7/441080/1080 [==============================] - 5s 5ms/step - loss: 0.3636 - acc: 0.8713Epoch 8/441080/1080 [==============================] - 5s 5ms/step - loss: 0.4806 - acc: 0.8704Epoch 9/441080/1080 [==============================] - 6s 5ms/step - loss: 0.5736 - acc: 0.8222Epoch 10/441080/1080 [==============================] - 5s 5ms/step - loss: 0.9006 - acc: 0.8065Epoch 11/441080/1080 [==============================] - 5s 5ms/step - loss: 1.2664 - acc: 0.7065Epoch 12/441080/1080 [==============================] - 5s 5ms/step - loss: 1.0772 - acc: 0.7426Epoch 13/441080/1080 [==============================] - 5s 5ms/step - loss: 0.6108 - acc: 0.8093Epoch 14/441080/1080 [==============================] - 5s 5ms/step - loss: 0.5305 - acc: 0.8537Epoch 15/441080/1080 [==============================] - 5s 5ms/step - loss: 0.3256 - acc: 0.9102Epoch 16/441080/1080 [==============================] - 5s 5ms/step - loss: 0.1295 - acc: 0.9491Epoch 17/441080/1080 [==============================] - 5s 5ms/step - loss: 0.0994 - acc: 0.9676Epoch 18/441080/1080 [==============================] - 5s 5ms/step - loss: 0.1509 - acc: 0.9639Epoch 19/441080/1080 [==============================] - 5s 5ms/step - loss: 0.2855 - acc: 0.8889Epoch 20/441080/1080 [==============================] - 5s 5ms/step - loss: 0.1235 - acc: 0.9620Epoch 21/441080/1080 [==============================] - 5s 5ms/step - loss: 0.0788 - acc: 0.9759Epoch 22/441080/1080 [==============================] - 5s 5ms/step - loss: 0.0639 - acc: 0.9769Epoch 23/441080/1080 [==============================] - 6s 5ms/step - loss: 0.0678 - acc: 0.9824Epoch 24/441080/1080 [==============================] - 5s 5ms/step - loss: 0.0255 - acc: 0.9917Epoch 25/441080/1080 [==============================] - 5s 5ms/step - loss: 0.0320 - acc: 0.9917Epoch 26/441080/1080 [==============================] - 5s 5ms/step - loss: 0.0136 - acc: 0.9954Epoch 27/441080/1080 [==============================] - 5s 5ms/step - loss: 0.0370 - acc: 0.9963Epoch 28/441080/1080 [==============================] - 5s 5ms/step - loss: 0.6671 - acc: 0.8111Epoch 29/441080/1080 [==============================] - 5s 5ms/step - loss: 0.5925 - acc: 0.8611Epoch 30/441080/1080 [==============================] - 5s 5ms/step - loss: 0.9083 - acc: 0.8028Epoch 31/441080/1080 [==============================] - 5s 5ms/step - loss: 0.7969 - acc: 0.7306Epoch 32/441080/1080 [==============================] - 5s 5ms/step - loss: 0.3669 - acc: 0.8676Epoch 33/441080/1080 [==============================] - 5s 5ms/step - loss: 0.2057 - acc: 0.9352Epoch 34/441080/1080 [==============================] - 5s 5ms/step - loss: 0.1457 - acc: 0.9528Epoch 35/441080/1080 [==============================] - 6s 5ms/step - loss: 0.1085 - acc: 0.9657Epoch 36/441080/1080 [==============================] - 5s 5ms/step - loss: 0.0829 - acc: 0.9694Epoch 37/441080/1080 [==============================] - 5s 5ms/step - loss: 0.1326 - acc: 0.9593Epoch 38/441080/1080 [==============================] - 5s 5ms/step - loss: 0.0626 - acc: 0.9750Epoch 39/441080/1080 [==============================] - 5s 5ms/step - loss: 0.0381 - acc: 0.9880Epoch 40/441080/1080 [==============================] - 5s 5ms/step - loss: 0.0153 - acc: 0.9963Epoch 41/441080/1080 [==============================] - 5s 5ms/step - loss: 0.1065 - acc: 0.9657Epoch 42/441080/1080 [==============================] - 5s 5ms/step - loss: 0.1160 - acc: 0.9713Epoch 43/441080/1080 [==============================] - 5s 5ms/step - loss: 0.0581 - acc: 0.9880Epoch 44/441080/1080 [==============================] - 5s 5ms/step - loss: 0.0293 - acc: 0.9926120/120 [==============================] - 0s 1ms/stepLoss = 0.0633144895857Test Accuracy = 0.983333333333