[Kaggle] dogs-vs-cats之模型训练

上一步建立好模型之后，现在就可以训练模型了。

主要代码如下：

import sys
#将当期路径加入系统path中
sys.path.append("E:\\CODE\\Anaconda\\tensorflow\\Kaggle\\My-TensorFlow-tutorials-master\\01 cats vs dogs\\")

import os
import numpy as np
import tensorflow as tf
import input_data
import model

#%%

N_CLASSES = 2 #类别数
IMG_W = 208  # resize the image, if the input image is too large, training will be very slow.
IMG_H = 208
BATCH_SIZE = 16
CAPACITY = 2000 #队列中元素个数
MAX_STEP = 10000 #最大迭代次数 with current parameters, it is suggested to use MAX_STEP>10k
learning_rate = 0.0001 # with current parameters, it is suggested to use learning rate<0.0001

#%%
def run_training():

    # you need to change the directories to yours.
    #train_dir = '/home/kevin/tensorflow/cats_vs_dogs/data/train/'#数据存放路径
    train_dir = 'E:\\data\\Dog_Cat\\train\\'
    #logs_train_dir = '/home/kevin/tensorflow/cats_vs_dogs/logs/train/'#存放训练参数，模型等
    logs_train_dir = "E:\\CODE\\Anaconda\\tensorflow\\Kaggle\\My-TensorFlow-tutorials-master\\01 cats vs dogs\\"

    train, train_label = input_data.get_files(train_dir)

    train_batch, train_label_batch = input_data.get_batch(train,
                                                          train_label,
                                                          IMG_W,
                                                          IMG_H,
                                                          BATCH_SIZE,
                                                          CAPACITY)
    train_logits = model.inference(train_batch, BATCH_SIZE, N_CLASSES)#获得模型的输出
    train_loss = model.losses(train_logits, train_label_batch)#获取loss
    train_op = model.trainning(train_loss, learning_rate)#训练模型
    train__acc = model.evaluation(train_logits, train_label_batch)#模型评估

    summary_op = tf.summary.merge_all()
    sess = tf.Session()
    train_writer = tf.summary.FileWriter(logs_train_dir, sess.graph)#把summary保存到路径中
    saver = tf.train.Saver()

    sess.run(tf.global_variables_initializer())
    coord = tf.train.Coordinator()
    threads = tf.train.start_queue_runners(sess=sess, coord=coord)

    try:
        for step in np.arange(MAX_STEP):
            if coord.should_stop():
                    break
            _, tra_loss, tra_acc = sess.run([train_op, train_loss, train__acc])

            if step % 50 == 0:
                print('Step %d, train loss = %.2f, train accuracy = %.2f%%' %(step, tra_loss, tra_acc*100.0))
                summary_str = sess.run(summary_op)
                train_writer.add_summary(summary_str, step)

            if step % 2000 == 0 or (step + 1) == MAX_STEP:
                checkpoint_path = os.path.join(logs_train_dir, 'model.ckpt')
                saver.save(sess, checkpoint_path, global_step=step)#保存模型及参数

    except tf.errors.OutOfRangeError:
        print('Done training -- epoch limit reached')
    finally:
        coord.request_stop()

    coord.join(threads)
    sess.close()

run_training()

一些函数说明如下：

1）tf.summary.merge_all

作用：Merges all summaries collected in the default graph.

2）tf.summary.FileWriter

作用：Writes Summary protocol buffers to event files.

3）tf.train.Saver

作用：保存和恢复变量。

举例：

saver.save(sess, 'my-model', global_step=0)

==> filename: 'my-model-0'
...
saver.save(sess, 'my-model', global_step=1000)

==> filename: 'my-model-1000' 

4）add_summary

作用：Writes Summary protocol buffers to event files.

程序运行后，控制台输出如下：

训练期间，也可以使用tensorboard查看模型训练情况。

可以使用如下命令打开tensorboard。

tensorboard --logdir=log文件路径

log文件路径即为程序中设置的logs_train_dir。

启动tensorboard之后，打开浏览器，输入对应网址，即可查看训练情况。

整体解码如下图：

loss与step的关系如下（两条曲线的原因是训练了两次，一次迭代了10000步，另一次迭代了15000步）：

也可以选择查看模型：

说明：

代码来自：https://github.com/kevin28520/My-TensorFlow-tutorials，略有修改

函数作用主要参考tensorflow官网。https://www.tensorflow.org/versions/master/api_docs/