Tensorflow学习记录10--tensorboard的使用

1 简介

tensorboard可以追踪loss以及accuracy的变化，追踪参数值w以及b的变化，以及可以显示卷积过程中的图像等等。

2 使用方式

1 首先标记要记录的参数

图像,值以及变量:

#图像
tf.image_summary(tag, tensor, max_images=3, collections=None, name=None)
#值
tf.scalar_summary(tags, values, collections=None, name=None)
#变量

例如:

#图像
tf.image_summary("x", x_new, max_images=1)
#值
cost_summary = tf.scalar_summary(cost.op.name,cost)

2 定义统计的op以及定义写入操作

# Merge all summaries into a single op
merged_summary_op = tf.merge_all_summaries()
# op to write logs to Tensorboard
summary_writer = tf.train.SummaryWriter("./log/",graph=tf.get_default_graph())

3 再会话sess中喂入数据得到统计的op的值，然后用写入操作把op的值用写入操作summary_writer写入指定目录即可。

    summary = sess.run(merged_summar_op,feed_dict={x:batch_x})
    summary_writer.add_summary(summary)

3 实例

1 图像

显示mnist图像

import numpy as np
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data

mnist = input_data.read_data_sets("/tmp/data/", one_hot=True)
x = tf.placeholder(tf.float32, [None, 784])
if __name__ == '__main__':

    with tf.Session() as sess:
        x_new = tf.reshape(x, shape=[-1, 28, 28, 1])
        tf.image_summary("x", x_new, max_images=1)
        # Merge all summaries into a single op
        merged_summary_op = tf.merge_all_summaries()
        # op to write logs to Tensorboard
        summary_writer = tf.train.SummaryWriter("./log/",graph=tf.get_default_graph())
        choose = np.random.randint(len(mnist.test.images))
        batch_x = mnist.test.images[choose].reshape([-1, 784])

        summary = sess.run(merged_summary_op,feed_dict={x: batch_x})
        summary_writer.add_summary(summary)

显示自定义图像

import numpy as np
import tensorflow as tf
x = tf.placeholder(tf.float32, [None, 784])
batch_x = np.random.randint(256,size=[1,784]).astype(np.uint8)
if __name__ == '__main__':

    with tf.Session() as sess:
        x_new = tf.reshape(x, shape=[-1, 28, 28, 1])
        tf.image_summary("x", x_new, max_images=1)
        # Merge all summaries into a single op
        merged_summary_op = tf.merge_all_summaries()
        # op to write logs to Tensorboard
        summary_writer = tf.train.SummaryWriter("./log/",graph=tf.get_default_graph())
        #choose = np.random.randint(len(mnist.test.images))
        #batch_x = mnist.test.images[choose].reshape([-1, 784])

        summary = sess.run(merged_summary_op,feed_dict={x: batch_x})
        summary_writer.add_summary(summary)

2 loss，精度等

代码

# -*- coding: utf-8 -*-
# 输入数据
import input_data
import pdb
mnist = input_data.read_data_sets("/tmp/data/", one_hot=True)

import tensorflow as tf

# 定义网络超参数
learning_rate = 0.001
training_iters = 200000
batch_size = 64
display_step = 20

# 定义网络参数
n_input = 784 # 输入的维度
n_classes = 10 # 标签的维度
dropout = 0.8 # Dropout 的概率

# 占位符输入
x = tf.placeholder(tf.float32, [None, n_input])
y = tf.placeholder(tf.float32, [None, n_classes])
keep_prob = tf.placeholder(tf.float32)

# 卷积操作
def conv2d(name, l_input, w, b):
    return tf.nn.relu(tf.nn.bias_add(tf.nn.conv2d(l_input, w, strides=[1, 1, 1, 1], padding='SAME'),b), name=name)

# 最大下采样操作
def max_pool(name, l_input, k):
    return tf.nn.max_pool(l_input, ksize=[1, k, k, 1], strides=[1, k, k, 1], padding='SAME', name=name)

# 归一化操作
def norm(name, l_input, lsize=4):
    return tf.nn.lrn(l_input, lsize, bias=1.0, alpha=0.001 / 9.0, beta=0.75, name=name)

# 定义整个网络 
def alex_net(_X, _weights, _biases, _dropout):
    # 向量转为矩阵
    # 这个是把x的-1维即最后一维，即每一副图像从一维变为28*28*1维的图像
    _X = tf.reshape(_X, shape=[-1, 28, 28, 1])

    # 卷积层
    conv1 = conv2d('conv1', _X, _weights['wc1'], _biases['bc1'])
    # 下采样层
    pool1 = max_pool('pool1', conv1, k=2)
    # 归一化层
    norm1 = norm('norm1', pool1, lsize=4)
    # Dropout
    norm1 = tf.nn.dropout(norm1, _dropout)

    # 卷积
    conv2 = conv2d('conv2', norm1, _weights['wc2'], _biases['bc2'])
    # 下采样
    pool2 = max_pool('pool2', conv2, k=2)
    # 归一化
    norm2 = norm('norm2', pool2, lsize=4)
    # Dropout
    norm2 = tf.nn.dropout(norm2, _dropout)

    # 卷积
    conv3 = conv2d('conv3', norm2, _weights['wc3'], _biases['bc3'])
    # 下采样
    pool3 = max_pool('pool3', conv3, k=2)
    # 归一化
    norm3 = norm('norm3', pool3, lsize=4)
    # Dropout
    norm3 = tf.nn.dropout(norm3, _dropout)

    # 全连接层，先把特征图转为向量
    dense1 = tf.reshape(norm3, [-1, _weights['wd1'].get_shape().as_list()[0]]) 
    dense1 = tf.nn.relu(tf.matmul(dense1, _weights['wd1']) + _biases['bd1'], name='fc1') 
    # 全连接层
    dense2 = tf.nn.relu(tf.matmul(dense1, _weights['wd2']) + _biases['bd2'], name='fc2') # Relu activation

    # 网络输出层
    out = tf.matmul(dense2, _weights['out']) + _biases['out']
    return out

# 存储所有的网络参数
weights = {
    'wc1': tf.Variable(tf.random_normal([3, 3, 1, 64])),
    'wc2': tf.Variable(tf.random_normal([3, 3, 64, 128])),
    'wc3': tf.Variable(tf.random_normal([3, 3, 128, 256])),
    'wd1': tf.Variable(tf.random_normal([4096, 1024])),
    #'wd1': tf.Variable(tf.random_normal([4\*4\*256, 1024])),
    'wd2': tf.Variable(tf.random_normal([1024, 1024])),
    'out': tf.Variable(tf.random_normal([1024, 10]))
}
biases = {
    'bc1': tf.Variable(tf.random_normal([64])),
    'bc2': tf.Variable(tf.random_normal([128])),
    'bc3': tf.Variable(tf.random_normal([256])),
    'bd1': tf.Variable(tf.random_normal([1024])),
    'bd2': tf.Variable(tf.random_normal([1024])),
    'out': tf.Variable(tf.random_normal([n_classes]))
}

# 构建模型
pred = alex_net(x, weights, biases, keep_prob)

# 定义损失函数和学习步骤
cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(pred, y))
optimizer = tf.train.AdamOptimizer(learning_rate=learning_rate).minimize(cost)

# 测试网络
correct_pred = tf.equal(tf.argmax(pred,1), tf.argmax(y,1))
accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))

# 初始化所有的共享变量
init = tf.initialize_all_variables()

# add summary
cost_summary = tf.scalar_summary(cost.op.name,cost)
accuracy_summary = tf.scalar_summary(accuracy.op.name,accuracy)

# 开启一个训练
with tf.Session() as sess:
    sess.run(init)
    step = 1
    summary_op = tf.merge_summary([cost_summary,accuracy_summary])
    summary_writer = tf.train.SummaryWriter("./log/",sess.graph)
    # Keep training until reach max iterations
    #while step \* batch_size < training_iters:
    while step * batch_size < training_iters:
        batch_xs, batch_ys = mnist.train.next_batch(batch_size)
        # 获取批数据
        sess.run(optimizer, feed_dict={x: batch_xs, y: batch_ys, keep_prob: dropout})
        if step % display_step == 0:
            # 计算精度
            acc = sess.run(accuracy, feed_dict={x: batch_xs, y: batch_ys, keep_prob: 1.})
            # 计算损失值
            loss = sess.run(cost, feed_dict={x: batch_xs, y: batch_ys, keep_prob: 1.})
            #print "Iter " + str(step\*batch_size) + ", Minibatch Loss= " + "{:.6f}".format(loss) + ", Training Accuracy= " + "{:.5f}".format(acc)
            print "Iter " + str(step*batch_size) + ", Minibatch Loss= " + "{:.6f}".format(loss) + ", Training Accuracy= " + "{:.5f}".format(acc)
        summary_op_out = sess.run(summary_op, feed_dict={x: batch_xs, y: batch_ys, keep_prob: 1.})
        summary_writer.add_summary(summary_op_out,step)
        step += 1
    print "Optimization Finished!"
    # 计算测试精度
    print "Testing Accuracy:", sess.run(accuracy, feed_dict={x: mnist.test.images[:256], y: mnist.test.labels[:256], keep_prob: 1.})

3 综合应用（显示卷积过程的中间特性）

"""
A Convolutional Network implementation example using TensorFlow library.
This example is using the MNIST database of handwritten digits
(http://yann.lecun.com/exdb/mnist/)

Author: Aymeric Damien
Project: https://github.com/aymericdamien/TensorFlow-Examples/
"""

import shutil as sh

import numpy as np
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data

print("Downloading MNIST data ...")
mnist = input_data.read_data_sets("/tmp/data/", one_hot=True)

# Log directory path.
log_path = "./log/test"
model_path = "./models/cnn_28x28_model.npy"
# Remove logging path.
sh.rmtree(log_path, ignore_errors=True)

# Parameters
batch_size = 1

# Network Parameters
n_input = 784  # MNIST data input (img shape: 28*28)
n_classes = 10  # MNIST total classes (0-9 digits)

# tf Graph input
x = tf.placeholder(tf.float32, [None, n_input])
y = tf.placeholder(tf.float32, [None, n_classes])
keep_prob = 1.

print ("Loading pre-trained model ...")
restore_data = np.load(model_path).item()

# Store layers weight & bias
weights = {
    # 5x5 conv, 1 input, 32 outputs
    'wc1': tf.constant(restore_data["weights"]["wc1"]),
    # 5x5 conv, 32 inputs, 64 outputs
    'wc2': tf.constant(restore_data["weights"]["wc2"]),
    # fully connected, 7*7*64 inputs, 1024 outputs
    'wd1': tf.constant(restore_data["weights"]["wd1"]),
    # 1024 inputs, 10 outputs (class prediction)
    'out': tf.constant(restore_data["weights"]["out"])
}

biases = {
    'bc1': tf.constant(restore_data["biases"]["bc1"]),
    'bc2': tf.constant(restore_data["biases"]["bc2"]),
    'bd1': tf.constant(restore_data["biases"]["bd1"]),
    'out': tf.constant(restore_data["biases"]["out"])
}


def visualize_conv_layer(x, ix, iy, channels, cx=8):
    """
    Aggregate the feature maps to an image from the given tensor of a
    convolution layer.

    Reference:
    http://stackoverflow.com/questions/33802336/visualizing-output-of
    -convolutional-layer-in-tensorflow

    :param x:           The tensor of a convolution layer.
    :param ix:          The width.
    :param iy:          The height.
    :param channels:    The depth (channel number).
    :param cx:          The number of how many feature maps in a row.
    :return:            The aggregated feature map.
    """
    cy = channels / cx
    print("ix=%d, iy=%d, channels=%d, cx=%d, cy=%d" %
          (ix, iy, channels, cx, cy))

    # First slice off 1 image and remove the image dimension.
    img = tf.slice(x, [0, 0, 0, 0], [1, -1, -1, -1])
    img = tf.reshape(img, [iy, ix, channels])

    # Add a couple of pixels of zero padding around the image
    ix += 4
    iy += 4
    img = tf.image.resize_image_with_crop_or_pad(img, iy, ix)

    img = tf.reshape(img, [iy, ix, cy, cx])
    img = tf.transpose(img, perm=[2, 0, 3, 1])
    img = tf.reshape(img, [1, cy * iy, cx * ix, 1])

    return img


# Create some wrappers for simplicity
def conv2d(x, W, b, strides=1):
    # Conv2D wrapper, with bias and relu activation
    x = tf.nn.conv2d(x, W, strides=[1, strides, strides, 1], padding='SAME')
    x = tf.nn.bias_add(x, b)
    return tf.nn.relu(x)


def maxpool2d(x, k=2):
    # MaxPool2D wrapper
    return tf.nn.max_pool(x, ksize=[1, k, k, 1], strides=[1, k, k, 1],
                          padding='SAME')


# Create model
def conv_net(x, weights, biases):
    # Reshape input picture
    x = tf.reshape(x, shape=[-1, 28, 28, 1])
    tf.image_summary("x", x, max_images=batch_size)

    # Convolution Layer
    conv1 = conv2d(x, weights['wc1'], biases['bc1'])
    print ""
    print ("conv1=%s" % conv1)
    tf.image_summary("conv1",
                     visualize_conv_layer(conv1,
                                          conv1.get_shape().as_list()[1],
                                          conv1.get_shape().as_list()[2],
                                          conv1.get_shape().as_list()[3],
                                          8),
                     max_images=1)

    # Max Pooling (down-sampling)
    conv1 = maxpool2d(conv1, k=2)
    print ("maxpool2d(conv1)=%s" % conv1)
    tf.image_summary("maxpool2d(conv1)",
                     visualize_conv_layer(conv1,
                                          conv1.get_shape().as_list()[1],
                                          conv1.get_shape().as_list()[2],
                                          conv1.get_shape().as_list()[3],
                                          8),
                     max_images=1)

    # Convolution Layer
    conv2 = conv2d(conv1, weights['wc2'], biases['bc2'])
    print ""
    print ("conv2=%s" % conv2)
    tf.image_summary("conv2",
                     visualize_conv_layer(conv2,
                                          conv2.get_shape().as_list()[1],
                                          conv2.get_shape().as_list()[2],
                                          conv2.get_shape().as_list()[3],
                                          8),
                     max_images=1)

    # Max Pooling (down-sampling)
    conv2 = maxpool2d(conv2, k=2)
    print ("maxpool2d(conv2)=%s" % conv2)
    tf.image_summary("maxpool2d(conv2)",
                     visualize_conv_layer(conv2,
                                          conv2.get_shape().as_list()[1],
                                          conv2.get_shape().as_list()[2],
                                          conv2.get_shape().as_list()[3],
                                          8),
                     max_images=1)

    # Fully connected layer
    # Reshape conv2 output to fit fully connected layer input
    fc1 = tf.reshape(conv2,
                     [-1, weights['wd1'].get_shape().as_list()[0]])
    fc1 = tf.add(tf.matmul(fc1, weights['wd1']), biases['bd1'])
    fc1 = tf.nn.relu(fc1)

    # Output, class prediction
    out = tf.add(tf.matmul(fc1, weights['out']), biases['out'])
    out = tf.nn.softmax(out)
    print ""

    return out


# Construct model
pred = conv_net(x, weights, biases)

# Initializing the variables
init = tf.initialize_all_variables()

# Launch the graph
if __name__ == '__main__':
    with tf.Session() as sess:
        sess.run(init)

        # Merge all summaries into a single op
        merged_summary_op = tf.merge_all_summaries()
        # op to write logs to Tensorboard
        summary_writer = tf.train.SummaryWriter(log_path,
                                                graph=tf.get_default_graph())

        # Prepare the test data randomly.
        choose = np.random.randint(len(mnist.test.images))
        batch_x = mnist.test.images[choose].reshape([-1, 784])

        # Run the prediction.
        final_pred, summary = sess.run([pred, merged_summary_op],
                                       feed_dict={x: batch_x})
        print ("The outcome is %s" % final_pred)

        # Write logs at every iteration
        summary_writer.add_summary(summary)
        print ("Use \"tensorboard --logdir=./log\" to launch the TensorBoard.")

参考：
https://github.com/boyw165/tensorflow-vgg.git
这里有个vgg的可视化
https://github.com/woodrush/vgg-visualizer-tf