一：tensorboard简介 tensorboard是tensorflow的可视化工具，可以显示图、展示中间结果。详见：https://tensorflow.google.cn/get_started/summaries_and_tensorboard 二：tf.sumary.histogram tf.sumary.histogram：直方图。 TensorBoard 直方图信息中心用于显示在 TensorFlow 图中某些 Tensor 随着时间推移而变化的分布。即，该信息中心可显示在不同时间点对应张量的许多张直方图图示。 tf.summary.histogram 基于任意大小和形状的张量，并将张量压缩成一个由许多分箱组成的直方图数据结构，这些分箱有各自的宽度和计数。
import tensorflow as tf k = tf.placeholder(tf.float32) # Make a normal distribution, with a shifting mean mean_moving_normal = tf.random_normal(shape=[1000], mean=(5*k), stddev=1) # Record that distribution into a histogram summary tf.summary.histogram("normal/moving_mean", mean_moving_normal) # Setup a session and summary writer sess = tf.Session() writer = tf.summary.FileWriter("/tmp/histogram_example") #writer = tf.summary.FileWriter("/Users/dudu/Desktop/studykeras") summaries = tf.summary.merge_all() # Setup a loop and write the summaries to disk N = 400 for step in range(N): k_val = step/float(N) summ = sess.run(summaries, feed_dict={k: k_val}) writer.add_summary(summ, global_step=step)执行后使用以下命令： tensorboard –logdir=/tmp/histogram_example tensorboard –logdir=/Users/dudu/Desktop/studykeras 三：tf.summary.scalar 一般在画loss,accuary时会用到这个函数。
import tensorflow as tf import numpy as np ## prepare the original data with tf.name_scope('data'): x_data = np.random.rand(100).astype(np.float32) y_data = 0.3*x_data+0.1 ##creat parameters with tf.name_scope('parameters'): with tf.name_scope('weights'): weight = tf.Variable(tf.random_uniform([1],-1.0,1.0)) tf.summary.histogram('weight',weight) with tf.name_scope('biases'): bias = tf.Variable(tf.zeros([1])) tf.summary.histogram('bias',bias) ##get y_prediction with tf.name_scope('y_prediction'): y_prediction = weight*x_data+bias ##compute the loss with tf.name_scope('loss'): loss = tf.reduce_mean(tf.square(y_data-y_prediction)) tf.summary.scalar('loss',loss) ##creat optimizer optimizer = tf.train.GradientDescentOptimizer(0.5) #creat train ,minimize the loss with tf.name_scope('train'): train = optimizer.minimize(loss) #creat init with tf.name_scope('init'): init = tf.global_variables_initializer() ##creat a Session sess = tf.Session() #merged merged = tf.summary.merge_all() ##initialize writer = tf.summary.FileWriter("/Users/dudu/Desktop/studykeras", sess.graph) sess.run(init) ## Loop for step in range(101): sess.run(train) rs=sess.run(merged) writer.add_summary(rs, step)执行后使用以下命令： tensorboard –logdir=/Users/dudu/Desktop/studykeras 四：tf.summary.image
import tensorflow as tf # 获取图片数据 file = open('/Users/dudu/Desktop/studykeras/4.png', 'rb') data = file.read() file.close() # 图片处理 image = tf.image.decode_png(data, channels=3) image = tf.expand_dims(image, 0) # 添加到日志中 sess = tf.Session() writer = tf.summary.FileWriter('/Users/dudu/Desktop/studykeras') summary_op = tf.summary.image("image1", image) # 运行并写入日志 summary = sess.run(summary_op) writer.add_summary(summary) # 关闭 writer.close() sess.close()将图片保存为png格式 执行后使用以下命令： tensorboard –logdir=/Users/dudu/Desktop/studykeras
本文参考了以下博客： https://www.cnblogs.com/fydeblog/p/7429344.html https://tensorflow.google.cn/programmers_guide/summaries_and_tensorboard https://blog.csdn.net/smf0504/article/details/56369758 等等
