TensorFlow_MNIST 保存、恢复模型及参数

一时失言乱红尘 2022-06-12 09:14 258阅读 0赞

内容：使用TensorFlow跑MNIST，并保存模型。之后恢复模型并进行测试  
配置：win7x64/PyCharm/Python3.5/tensorflow-1.2.1/dataSet-MNIST  
原程序：[https://github.com/tensorflow/tensorflow/tree/master/tensorflow/examples/tutorials/mnist][https_github.com_tensorflow_tensorflow_tree_master_tensorflow_examples_tutorials_mnist]  
![这里写图片描述][SouthEast]

**1.网络定义**

def deepnn(x):
      """deepnn builds the graph for a deep net for classifying digits. Args: x: an input tensor with the dimensions (N_examples, 784), where 784 is the number of pixels in a standard MNIST image. Returns: A tuple (y, keep_prob). y is a tensor of shape (N_examples, 10), with values equal to the logits of classifying the digit into one of 10 classes (the digits 0-9). keep_prob is a scalar placeholder for the probability of dropout. """
      # Reshape to use within a convolutional neural net.
      # Last dimension is for "features" - there is only one here, since images are
      # grayscale -- it would be 3 for an RGB image, 4 for RGBA, etc.
      with tf.name_scope('reshape'):
        x_image = tf.reshape(x, [-1, 28, 28, 1])   # [batch,height,width,channels]
    
      # First convolutional layer - maps one grayscale image to 32 feature maps.
      with tf.name_scope('conv1'):
        W_conv1 = weight_variable([5, 5, 1, 32])   # shape of weight [conv_W,conv_H,channal_before_conv,channal_after_conv]
        b_conv1 = bias_variable([32])
        h_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) + b_conv1)
    
      # Pooling layer - downsamples by 2X.
      with tf.name_scope('pool1'):
        h_pool1 = max_pool_2x2(h_conv1)
    
      # Second convolutional layer -- maps 32 feature maps to 64.
      with tf.name_scope('conv2'):
        W_conv2 = weight_variable([5, 5, 32, 64])
        b_conv2 = bias_variable([64])
        h_conv2 = tf.nn.relu(conv2d(h_pool1, W_conv2) + b_conv2)
    
      # Second pooling layer.
      with tf.name_scope('pool2'):
        h_pool2 = max_pool_2x2(h_conv2)
    
      # Fully connected layer 1 -- after 2 round of downsampling, our 28x28 image
      # is down to 7x7x64 feature maps -- maps this to 1024 features.
      with tf.name_scope('fc1'):
        W_fc1 = weight_variable([7 * 7 * 64, 1024])
        b_fc1 = bias_variable([1024])
    
        h_pool2_flat = tf.reshape(h_pool2, [-1, 7*7*64])
        h_fc1 = tf.nn.relu(tf.matmul(h_pool2_flat, W_fc1) + b_fc1)
    
      # Dropout - controls the complexity of the model, prevents co-adaptation of
      # features.
      with tf.name_scope('dropout'):
        keep_prob = tf.placeholder(tf.float32,name='keep_prob')
        h_fc1_drop = tf.nn.dropout(h_fc1, keep_prob)
    
      # Map the 1024 features to 10 classes, one for each digit
      with tf.name_scope('fc2'):
        W_fc2 = weight_variable([1024, 10])
        b_fc2 = bias_variable([10])
    
        y_conv = tf.add(tf.matmul(h_fc1_drop, W_fc2), b_fc2, name='output')
      return y_conv, keep_prob
    
    
    def conv2d(x, W):
      """conv2d returns a 2d convolution layer with full stride."""
      return tf.nn.conv2d(x, W, strides=[1, 1, 1, 1], padding='SAME')
    
    
    def max_pool_2x2(x):
      """max_pool_2x2 downsamples a feature map by 2X."""
      return tf.nn.max_pool(x, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
    
    
    def weight_variable(shape):
      """weight_variable generates a weight variable of a given shape."""
      initial = tf.truncated_normal(shape, stddev=0.1)
      return tf.Variable(initial)
    
    
    def bias_variable(shape):
      """bias_variable generates a bias variable of a given shape."""
      initial = tf.constant(0.1, shape=shape)
      return tf.Variable(initial)

**2.训练，保存模型及参数**

# Import data
      mnist = input_data.read_data_sets(FLAGS.data_dir, one_hot=True)
     # Create the model
      x = tf.placeholder(tf.float32, [None, 784], name='input_x')
    
      # Define loss and optimizer
      y_ = tf.placeholder(tf.float32, [None, 10], name='input_y_')
    
      # Build the graph for the deep net
      y_conv, keep_prob = deepnn(x)
    
      with tf.name_scope('loss'):
        cross_entropy = tf.nn.softmax_cross_entropy_with_logits(labels=y_,
                                                                logits=y_conv)
        cross_entropy = tf.reduce_mean(cross_entropy)
    
      with tf.name_scope('adam_optimizer'):
        train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)
    
      with tf.name_scope('accuracy'):
        correct_prediction = tf.equal(tf.argmax(y_conv, 1), tf.argmax(y_, 1))
        # argmax: return the index with max value among 10 values
        correct_prediction = tf.cast(correct_prediction, tf.float32)
        accuracy = tf.reduce_mean(correct_prediction)
    
    
      graph_location = tempfile.mkdtemp() # temp file
      print('Saving graph to: %s' % graph_location)
      train_writer = tf.summary.FileWriter(graph_location)
      train_writer.add_graph(tf.get_default_graph())
      saver=tf.train.Saver(tf.global_variables())
    
    
      with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())
        # train
        for i in range(300):  #20000
          batch = mnist.train.next_batch(50)
          if i % 100 == 0:
            train_accuracy = accuracy.eval(feed_dict={
                x: batch[0], y_: batch[1], keep_prob: 1.0})
            print('step %d, training accuracy %g' % (i, train_accuracy))
    
            #save variables
            save_path ='G:/tf_model/model_'+'%d'%i
            print('%s' % save_path)
            saver.save(sess, save_path)
            # #save model
            # save_path = 'G:/tf_model'
            # save_name = 'test_' + '%d' % i + '.pb'
            # tf.train.write_graph(sess.graph_def, save_path, save_name, False) #save graph
            # saver.save(sess, save_path+'/checkpoint.data') #save data
          train_step.run(feed_dict={x: batch[0], y_: batch[1], keep_prob: 0.5})

保存的文件：  
![保存的文件][SouthEast 1]

**3.恢复模型及参数，测试**

# test 1 <using graph> -> load saved graph and variables
      sess1 = tf.Session()
      saver1 = tf.train.import_meta_graph('G:/tf_model/model_200.meta')  # load graph
      saver1.restore(sess1, 'G:/tf_model/model_200')  # load variables (remember no extension of file)
      # saver1.restore(sess1, tf.train.latest_checkpoint('G:/tf_model/')) # or this
      graph = tf.get_default_graph()  # get graph
    
      # get tensor by name, the name is named when define the network
      x = graph.get_tensor_by_name('input_x:0')    # input image
      y_ = graph.get_tensor_by_name('input_y_:0')  # input label
      y_conv = graph.get_tensor_by_name('fc2/output:0')  # output result from deepNN (predict label)
      keep_prob_ = graph.get_tensor_by_name('dropout/keep_prob:0')  # keep probability
    
      # accuracy calculation
      correct_prediction = tf.equal(tf.argmax(y_conv, 1), tf.argmax(y_, 1))  # compare predict result and label
      correct_prediction = tf.cast(correct_prediction, tf.float32)  # convert bool to float
      accuracy1 = tf.reduce_mean(correct_prediction)  # calculate mean accuracy
    
      for j in range(10):
        testSample_start = j * 50      # start num of test sample of one batch
        testSample_end = (j + 1) * 50  # end num of test sample of one batch <use batch as the capacity of GPU is limited>
        print('test %d accuracy %g' % (j, accuracy1.eval(session=sess1, feed_dict={ x: mnist.test.images[testSample_start:testSample_end], y_: mnist.test.labels[testSample_start:testSample_end], keep_prob_: 1.0})))  # feed data for test

测试结果（训练次数较少，测试结果准确性不高）  
![SouthEast 2][]

[https_github.com_tensorflow_tensorflow_tree_master_tensorflow_examples_tutorials_mnist]: https://github.com/tensorflow/tensorflow/tree/master/tensorflow/examples/tutorials/mnist
[SouthEast]: /images/20220612/26e9a7ab6e454773881e7d9804b3bf0a.png
[SouthEast 1]: /images/20220612/f380febbe72e461a8b38b9d5b9435fed.png
[SouthEast 2]: /images/20220612/97bb710ed48e470c8ae56695009b4929.png