cifar10数据集训练
有关CIFAR-10数据集
（1）CIFAR-10数据集由10个类的60000个32x32彩色图像组成，每个类有6000个图像。有50000个训 练图像和10000个测试图像。
（2）数据集分为五个训练批次和一个测试批次，每个批次有10000个图像。测试批次包含来自每个类别 的恰好1000个随机选择的图像。
（3）第一部分是特征部分，使用一个[10000,3072的uint8的矩阵进行存储，每一行向量都是3X3大小的 3通道图片，构成的格式类似于[3,3,3]；第二部分为标签部分，使用一个10000数据的list进行存 储，每个list对应的是0-9中的一个数字，对应于物品分类。另外对于python的数据集，还有一个 标签为“label_names”，例如label_names[0] == “airplane”等。
CIFAR-10数据集下载
官网：http://www.cs.toronto.edu/~kriz/cifar.html
CIFAR-10数据集训练
最终可以运行的代码和一些解释：
1. cifar10_input.py
"""Routine for decoding the CIFAR-10 binary file format."""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import os
from six.moves import xrange # pylint: disable=redefined-builtin
import tensorflow as tf
# Process images of this size. Note that this differs from the original CIFAR
# image size of 32 x 32. If one alters this number, then the entire model
# architecture will change and any model would need to be retrained.
# 原图像的尺度为32*32,但根据常识，信息部分通常位于图像的中央，
# 这里定义了以中心裁剪后图像的尺寸
IMAGE_SIZE = 24
# Global constants describing the CIFAR-10 data set.
NUM_CLASSES = 10
NUM_EXAMPLES_PER_EPOCH_FOR_TRAIN = 50000
NUM_EXAMPLES_PER_EPOCH_FOR_EVAL = 10000
def read_cifar10(filename_queue):
"""Reads and parses examples from CIFAR10 data files.
Recommendation: if you want N-way read parallelism, call this function
N times. This will give you N independent Readers reading different
files & positions within those files, which will give better mixing of
examples.
Args:
filename_queue: A queue of strings with the filenames to read from.
Returns:
An object representing a single example, with the following fields:
height: number of rows in the result (32)
width: number of columns in the result (32)
depth: number of color channels in the result (3)
key: a scalar string Tensor describing the filename & record number
for this example.
label: an int32 Tensor with the label in the range 0..9.
uint8image: a [height, width, depth] uint8 Tensor with the image data
"""
# 定义一个空的类对象，类似于c语言里面的结构体定义
class CIFAR10Record(object):
pass
result = CIFAR10Record()
# Dimensions of the images in the CIFAR-10 dataset.
# See http://www.cs.toronto.edu/~kriz/cifar.html for a description of the
# input format.
label_bytes = 1 # 2 for CIFAR-100
result.height = 32
result.width = 32
result.depth = 3
# 一张图像占用空间
image_bytes = result.height * result.width * result.depth
# Every record consists of a label followed by the image, with a
# fixed number of bytes for each.
# 数据集中一条记录的组成
record_bytes = label_bytes + image_bytes
# Read a record, getting filenames from the filename_queue. No
# header or footer in the CIFAR-10 format, so we leave header_bytes
# and footer_bytes at their default of 0.
# 定义一个Reader，它每次能从文件中读取固定字节数
reader = tf.FixedLengthRecordReader(record_bytes=record_bytes)
# 返回从filename_queue中读取的(key, value)对，key和value都是字符串类型的tensor，并且当队列中的某一个文件读完成时，该文件名会dequeue
result.key, value = reader.read(filename_queue)
# Convert from a string to a vector of uint8 that is record_bytes long.
# 解码操作可以看作读二进制文件，把字符串中的字节转换为数值向量,每一个数值占用一个字节,在[0, 255]区间内，因此out_type要取uint8类型
record_bytes = tf.decode_raw(value, tf.uint8) # 将字符串Tensor转化成uint8类型
# The first bytes represent the label, which we convert from uint8->int32.
# 从一维tensor对象中截取一个slice,类似于从一维向量中筛选子向量，因为record_bytes中包含了label和feature，故要对向量类型tensor进行'parse'操作
result.label = tf.cast(
tf.strided_slice(record_bytes, [0], [label_bytes]), tf.int32) # 分别表示待截取片段的起点和长度，并且把标签由之前的uint8转变成int32数据类型
# The remaining bytes after the label represent the image, which we reshape.
# from [depth * height * width] to [depth, height, width].
# 提取每条记录中的图像数据为result.depth, result.height, result.width
depth_major = tf.reshape(
tf.strided_slice(record_bytes, [label_bytes],
[label_bytes + image_bytes]),
[result.depth, result.height, result.width])
# Convert from [depth, height, width] to [height, width, depth].
# 改变为height, width, depth
result.uint8image = tf.transpose(depth_major, [1, 2, 0])
return result
# 构建一个排列后的一组图片和分类
def _generate_image_and_label_batch(image, label, min_queue_examples,
batch_size, shuffle):
"""Construct a queued batch of images and labels.
Args:
image: 3-D Tensor of [height, width, 3] of type.float32.
label: 1-D Tensor of type.int32
min_queue_examples: int32, minimum number of samples to retain
in the queue that provides of batches of examples.
batch_size: Number of images per batch.
shuffle: boolean indicating whether to use a shuffling queue.
Returns:
images: Images. 4D tensor of [batch_size, height, width, 3] size.
labels: Labels. 1D tensor of [batch_size] size.
"""
# Create a queue that shuffles the examples, and then
# read 'batch_size' images + labels from the example queue.
# 线程数
num_preprocess_threads = 16
# 布尔指示是否使用一个shuffling队列
if shuffle:
images, label_batch = tf.train.shuffle_batch(
[image, label],
batch_size=batch_size,
num_threads=num_preprocess_threads,
capacity=min_queue_examples + 3 * batch_si***_after_dequeue=min_queue_examples)
else:
# tf.train.batch(tensors, batch_size, num_threads=1, capacity=32,
# enqueue_many=False, shapes=None, dynamic_pad=False,
# allow_smaller_final_batch=False, shared_name=None, name=None)
# 这里是用队列实现，已经默认使用enqueue_runner将enqueue_runner加入到Graph'senqueue_runner集合中
# 其默认enqueue_many=False时，输入的tensor为一个样本【x,y,z】,输出为Tensor的一批样本
# capacity：队列中允许最大元素个数
images, label_batch = tf.train.batch(
[image, label],
batch_size=batch_size,
num_threads=num_preprocess_threads,
capacity=min_queue_examples + 3 * batch_size)
# Display the training images in the visualizer.
# 将训练图片可视化，可拱直接检查图片正误
tf.summary.image('images', images)
return images, tf.reshape(label_batch, [batch_size])
# 为CIFAR评价构建输入
# data_dir路径
# batch_size一个组的大小
def distorted_inputs(data_dir, batch_size):
"""Construct distorted input for CIFAR training using the Reader ops.
Args:
data_dir: Path to the CIFAR-10 data directory.
batch_size: Number of images per batch.
Returns:
images: Images. 4D tensor of [batch_size, IMAGE_SIZE, IMAGE_SIZE, 3] size.
labels: Labels. 1D tensor of [batch_size] size.
"""
filenames = [os.path.join(data_dir, 'data_batch_%d.bin' % i)
for i in xrange(1, 6)]
for f in filenames:
if not tf.gfile.Exists(f):
raise ValueError('Failed to find file: ' + f)
# Create a queue that produces the filenames to read.
filename_queue = tf.train.string_input_producer(filenames)
# Read examples from files in the filename queue.
read_input = read_cifar10(filename_queue)
reshaped_image = tf.cast(read_input.uint8image, tf.float32)
height = IMAGE_SIZE
width = IMAGE_SIZE
# Image processing for training the network. Note the many random
# distortions applied to the image.
# Randomly crop a [height, width] section of the image.
distorted_image = tf.random_crop(reshaped_image, [height, width, 3])
# Randomly flip the image horizontally.
distorted_image = tf.image.random_flip_left_right(distorted_image)
# Because these operations are not commutative, consider randomizing
# the order their operation.
# NOTE: since per_image_standardization zeros the mean and makes
# the stddev unit, this likely has no effect see tensorflow#1458.
distorted_image = tf.image.random_brightness(distorted_image,
max_delta=63)
distorted_image = tf.image.random_contrast(distorted_image,
lower=0.2, upper=1.8)
# Subtract off the mean and divide by the variance of the pixels.
float_image = tf.image.per_image_standardization(distorted_image)
# Set the shapes of tensors.
# 设置张量的型
float_image.set_shape([height, width, 3])
read_input.label.set_shape([1])
# Ensure that the random shuffling has good mixing properties.
# 确保洗牌的随机性
min_fraction_of_examples_in_queue = 0.4
min_queue_examples = int(NUM_EXAMPLES_PER_EPOCH_FOR_TRAIN *
min_fraction_of_examples_in_queue)
print('Filling queue with %d CIFAR images before starting to train. '
'This will take a few minutes.' % min_queue_examples)
# Generate a batch of images and labels by building up a queue of examples.
return _generate_image_and_label_batch(float_image, read_input.label,
min_queue_examples, batch_size,
shuffle=True)
# 为CIFAR评价构建输入
# eval_data使用训练还是评价数据集
# data_dir路径
# batch_size一个组的大小
def inputs(eval_data, data_dir, batch_size):
"""Construct input for CIFAR evaluation using the Reader ops.
Args:
eval_data: bool, indicating if one should use the train or eval data set.
data_dir: Path to the CIFAR-10 data directory.
batch_size: Number of images per batch.
Returns:
images: Images. 4D tensor of [batch_size, IMAGE_SIZE, IMAGE_SIZE, 3] size.
labels: Labels. 1D tensor of [batch_size] size.
"""
if not eval_data:
filenames = [os.path.join(data_dir, 'data_batch_%d.bin' % i)
for i in xrange(1, 6)]
num_examples_per_epoch = NUM_EXAMPLES_PER_EPOCH_FOR_TRAIN
else:
filenames = [os.path.join(data_dir, 'test_batch.bin')]
num_examples_per_epoch = NUM_EXAMPLES_PER_EPOCH_FOR_EVAL
for f in filenames:
if not tf.gfile.Exists(f):
raise ValueError('Failed to find file: ' + f)
# Create a queue that produces the filenames to read.
# 文件名队列
# def string_input_producer(string_tensor,
# num_epochs=None,
# shuffle=True,
# seed=None,
# capacity=32,
# shared_name=None,
# name=None,
# cancel_op=None):
# 根据上面的函数可以看出下面的这个默认对输入队列进行shuffle，string_input_producer返回的是字符串队列，
# 使用enqueue_runner将enqueue_runner加入到Graph'senqueue_runner集合中
filename_queue = tf.train.string_input_producer(filenames)
# Read examples from files in the filename queue.
# 从文件队列中读取解析出的图片队列
# read_cifar10从输入文件名队列中读取一条图像记录
read_input = read_cifar10(filename_queue)
# 将记录中的图像记录转换为float32
reshaped_image = tf.cast(read_input.uint8image, tf.float32)
height = IMAGE_SIZE
width = IMAGE_SIZE
# Image processing for evaluation.
# Crop the central [height, width] of the image.
# 将图像裁剪成24*24
resized_image = tf.image.resize_image_with_crop_or_pad(reshaped_image,
height, width)
# Subtract off the mean and divide by the variance of the pixels.
# 对图像数据进行归一化
float_image = tf.image.per_image_standardization(resized_image)
# Set the shapes of tensors.
float_image.set_shape([height, width, 3])
read_input.label.set_shape([1])
# Ensure that the random shuffling has good mixing properties.
min_fraction_of_examples_in_queue = 0.4
min_queue_examples = int(num_examples_per_epoch *
min_fraction_of_examples_in_queue)
# Generate a batch of images and labels by building up a queue of examples.
# 根据当前记录中第一条记录的值，采用多线程的方法，批量读取一个batch中的数据
return _generate_image_and_label_batch(float_image, read_input.label,
min_queue_examples, batch_size,
shuffle=False)
2. cifar10.py
"""Builds the CIFAR-10 network.
Summary of available functions:
# Compute input images and labels for training. If you would like to run
# evaluations, use inputs() instead.
inputs, labels = distorted_inputs()
# Compute inference on the model inputs to make a prediction.
predictions = inference(inputs)
# Compute the total loss of the prediction with respect to the labels.
loss = loss(predictions, labels)
# Create a graph to run one step of training with respect to the loss.
train_op = train(loss, global_step)
"""
# pylint: disable=missing-docstring
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import argparse
import os
import re
import sys
import tarfile
import argparse
from six.moves import urllib
import tensorflow as tf
import cifar10_input
parser = argparse.ArgumentParser()
# Basic model parameters.
parser.add_argument('--batch_size', type=int, default=128,
help='Number of images to process in a batch.')
parser.add_argument('--data_dir', type=str, default='D:/QQ文件/cifar10-test1/cifar10_data/',
help='Path to the CIFAR-10 data directory.')
parser.add_argument('--use_fp16', type=bool, default=False,
help='Train the model using fp16.')
FLAGS = parser.parse_args()
# Global constants describing the CIFAR-10 data set.
IMAGE_SIZE = cifar10_input.IMAGE_SIZE
NUM_CLASSES = cifar10_input.NUM_CLASSES
NUM_EXAMPLES_PER_EPOCH_FOR_TRAIN = cifar10_input.NUM_EXAMPLES_PER_EPOCH_FOR_TRAIN
NUM_EXAMPLES_PER_EPOCH_FOR_EVAL = cifar10_input.NUM_EXAMPLES_PER_EPOCH_FOR_EVAL
# Constants describing the training process.
MOVING_AVERAGE_DECAY = 0.9999 # The decay to use for the moving average.
NUM_EPOCHS_PER_DECAY = 350.0 # Epochs after which learning rate decays.# 衰减呈阶梯函数，控制衰减周期（阶梯宽度）
LEARNING_RATE_DECAY_FACTOR = 0.1 # Learning rate decay factor.# 学习率衰减因子
INITIAL_LEARNING_RATE = 0.1 # Initial learning rate.# 初始学习率
# If a model is trained with multiple GPUs, prefix all Op names with tower_name
# to differentiate the operations. Note that this prefix is removed from the
# names of the summaries when visualizing a model.
TOWER_NAME = 'tower'
DATA_URL = 'https://www.cs.toronto.edu/~kriz/cifar-10-binary.tar.gz'
def _activation_summary(x):
"""Helper to create summaries for activations.
Creates a summary that provides a histogram of activations.
Creates a summary that measures the sparsity of activations.
Args:
x: Tensor
Returns:
nothing
"""
# Remove 'tower_[0-9]/' from the name in case this is a multi-GPU training
# session. This helps the clarity of presentation on tensorboard.
tensor_name = re.sub('%s_[0-9]*/' % TOWER_NAME, '', x.op.name)
tf.summary.histogram(tensor_name + '/activations', x)
tf.summary.scalar(tensor_name + '/sparsity',
tf.nn.zero_fraction(x))
def _variable_on_cpu(name, shape, initializer):
"""Helper to create a Variable stored on CPU memory.
Args:
name: name of the variable
shape: list of ints
initializer: initializer for Variable
Returns:
Variable Tensor
"""
with tf.device('/cpu:0'): # 一个 context manager,用于为新的op指定要使用的硬件
dtype = tf.float16 if FLAGS.use_fp16 else tf.float32
var = tf.get_variable(name, shape, initializer=initializer, dtype=dtype)
return var
def _variable_with_weight_decay(name, shape, stddev, wd):
"""Helper to create an initialized Variable with weight decay.
Note that the Variable is initialized with a truncated normal distribution.
A weight decay is added only if one is specified.
Args:
name: name of the variable
shape: list of ints
stddev: standard deviation of a truncated Gaussian
wd: add L2Loss weight decay multiplied by this float. If None, weight
decay is not added for this Variable.
Returns:
Variable Tensor
"""
dtype = tf.float16 if FLAGS.use_fp16 else tf.float32
var = _variable_on_cpu(
name,
shape,
tf.truncated_normal_initializer(stddev=stddev, dtype=dtype))
if wd is not None:
weight_decay = tf.multiply(tf.nn.l2_loss(var), wd, name='weight_loss')
tf.add_to_collection('losses', weight_decay)
return var
def distorted_inputs():
"""Construct distorted input for CIFAR training using the Reader ops.
Returns:
images: Images. 4D tensor of [batch_size, IMAGE_SIZE, IMAGE_SIZE, 3] size.
labels: Labels. 1D tensor of [batch_size] size.
Raises:
ValueError: If no data_dir
"""
if not FLAGS.data_dir:
raise ValueError('Please supply a data_dir')
data_dir = os.path.join(FLAGS.data_dir, 'cifar-10-batches-bin')
images, labels = cifar10_input.distorted_inputs(data_dir=data_dir,
batch_size=FLAGS.batch_size)
if FLAGS.use_fp16:
images = tf.cast(images, tf.float16)
labels = tf.cast(labels, tf.float16)
return images, labels
def inputs(eval_data):
"""Construct input for CIFAR evaluation using the Reader ops.
Args:
eval_data: bool, indicating if one should use the train or eval data set.
Returns:
images: Images. 4D tensor of [batch_size, IMAGE_SIZE, IMAGE_SIZE, 3] size.
labels: Labels. 1D tensor of [batch_size] size.
Raises:
ValueError: If no data_dir
"""
if not FLAGS.data_dir:
raise ValueError('Please supply a data_dir')
data_dir = os.path.join(FLAGS.data_dir, 'cifar-10-batches-bin')
images, labels = cifar10_input.inputs(eval_data=eval_data,
data_dir=data_dir,
batch_size=FLAGS.batch_size)
if FLAGS.use_fp16:
images = tf.cast(images, tf.float16)
labels = tf.cast(labels, tf.float16)
return images, labels
# 开始建立网络，第一层卷积层的 weight 不进行 L2正则，因此 kernel(wd) 这一项设为0，建立值为0的 biases，
# conv1的结果由 ReLu 激活，由 _activation_summary() 进行汇总；然后建立第一层池化层，
# 最大池化尺寸和步长不一致可以增加数据的丰富性；最后建立 LRN 层
def inference(images):
"""Build the CIFAR-10 model.
Args:
images: Images returned from distorted_inputs() or inputs().
Returns:
Logits.
"""
# We instantiate all variables using tf.get_variable() instead of
# tf.Variable() in order to share variables across multiple GPU training runs.
# If we only ran this model on a single GPU, we could simplify this function
# by replacing all instances of tf.get_variable() with tf.Variable().
#
# conv1
with tf.variable_scope('conv1') as scope:
kernel = _variable_with_weight_decay('weights',
shape=[5, 5, 3, 64],
stddev=5e-2,
wd=0.0)
conv = tf.nn.conv2d(images, kernel, [1, 1, 1, 1], padding='SAME')
biases = _variable_on_cpu('biases', [64], tf.constant_initializer(0.0))
pre_activation = tf.nn.bias_add(conv, biases)
conv1 = tf.nn.relu(pre_activation, name=scope.name)
_activation_summary(conv1)
# pool1
pool1 = tf.nn.max_pool(conv1, ksize=[1, 3, 3, 1], strides=[1, 2, 2, 1],
padding='SAME', name='pool1')
# norm1 局部相响应归一化
# LRN层模仿了生物神经系统的
# "侧抑制"
# 机制，对局部神经元的活动创建竞争环境，使得其中响应比较大的值变得相对更大，并抑制其他反馈较小的神经元，增强了模型的泛化能力，LRN
# 对Relu 这种没有上限边界的激活函数会比较有用，因为它会从附近的多个卷积核的响应中挑选比较大的反馈，但不适合
# sigmoid这种有固定边界并且能抑制过大的激活函数。
norm1 = tf.nn.lrn(pool1, 4, bias=1.0, alpha=0.001 / 9.0, beta=0.75,
name='norm1')
# conv2
# 第二层卷积层与第一层，除了输入参数的改变之外，将 biases 值全部初始化为0.1，
# 调换最大池化和 LRN 层的顺序，先进行LRN，再使用最大池化层。
with tf.variable_scope('conv2') as scope:
kernel = _variable_with_weight_decay('weights',
shape=[5, 5, 64, 64],
stddev=5e-2,
wd=0.0)
conv = tf.nn.conv2d(norm1, kernel, [1, 1, 1, 1], padding='SAME')
biases = _variable_on_cpu('biases', [64], tf.constant_initializer(0.1))
pre_activation = tf.nn.bias_add(conv, biases)
conv2 = tf.nn.relu(pre_activation, name=scope.name)
_activation_summary(conv2)
# norm2
norm2 = tf.nn.lrn(conv2, 4, bias=1.0, alpha=0.001 / 9.0, beta=0.75,
name='norm2')
# pool2
pool2 = tf.nn.max_pool(norm2, ksize=[1, 3, 3, 1],
strides=[1, 2, 2, 1], padding='SAME', name='pool2')
# local3
# 第三层全连接层 ，需要先把前面的卷积层的输出结果全部 flatten，
# 使用 tf.reshape 函数将每个样本都变为一维向量，使用 get_shape 函数获取数据扁平化之后的长度；
# 然后对全连接层的 weights 和 biases 进行初始化，为了防止全连接层过拟合，设置一个非零的 wd 值0.004，
# 让这一层的所有参数都被 L2正则所约束，最后依然使用 Relu 激活函数进行非线性化。
# 同理，可以建立第四层全连接层。
with tf.variable_scope('local3') as scope:
# Move everything into depth so we can perform a single matrix multiply.
reshape = tf.reshape(pool2, [FLAGS.batch_size, -1])
dim = reshape.get_shape()[1].value
weights = _variable_with_weight_decay('weights', shape=[dim, 384],
stddev=0.04, wd=0.004)
biases = _variable_on_cpu('biases', [384], tf.constant_initializer(0.1))
local3 = tf.nn.relu(tf.matmul(reshape, weights) + biases, name=scope.name)
_activation_summary(local3)
# local4
with tf.variable_scope('local4') as scope:
weights = _variable_with_weight_decay('weights', shape=[384, 192],
stddev=0.04, wd=0.004)
biases = _variable_on_cpu('biases', [192], tf.constant_initializer(0.1))
local4 = tf.nn.relu(tf.matmul(local3, weights) + biases, name=scope.name)
_activation_summary(local4)
# linear layer(WX + b),
# We don't apply softmax here because
# tf.nn.sparse_softmax_cross_entropy_with_logits accepts the unscaled logits
# and performs the softmax internally for efficiency.
# 最后的 softmax_linear 层，先创建这一层的 weights 和 biases，不添加L2正则化。
# 在这个模型中，不像之前的例子使用 sotfmax 输出最后的结果，因为将 softmax 的操作放在来计算 loss 的部分，
# 将 softmax_linear 的线性返回值 logits 与 labels 计算 loss，
with tf.variable_scope('softmax_linear') as scope:
weights = _variable_with_weight_decay('weights', [192, NUM_CLASSES],
stddev=1 / 192.0, wd=0.0)
biases = _variable_on_cpu('biases', [NUM_CLASSES],
tf.constant_initializer(0.0))
softmax_linear = tf.add(tf.matmul(local4, weights), biases, name=scope.name)
_activation_summary(softmax_linear)
return softmax_linear
# 损失函数
# 通过 tf.nn.softmax 后的 logits 值(属于每个类别的概率值)
def loss(logits, labels):
"""Add L2Loss to all the trainable variables.
Add summary for "Loss" and "Loss/avg".
Args:
logits: Logits from inference().
labels: Labels from distorted_inputs or inputs(). 1-D tensor
of shape [batch_size]
Returns:
Loss tensor of type float.
"""
# Calculate the average cross entropy loss across the batch.
labels = tf.cast(labels, tf.int64)
# 在 CIFAR-10 中，labels 的 shape 为 [batch_size]，每个样本的 label 为0到9的一个数，代表10个分类，
# 这些类之间是相互排斥的，每个 CIFAR-10 图片只能被标记为唯一的一个标签：一张图片可能是一只狗或一辆卡车，而不能两者都是。
# 因此我们需要对 label 值 onehot encoding，转化过程比较繁琐，
# 新版的 TensorFlow API 支持对唯一值 labels 的 sparse_to_dense，只需要一步：
cross_entropy = tf.nn.sparse_softmax_cross_entropy_with_logits(
labels=labels, logits=logits, name='cross_entropy_per_example')
cross_entropy_mean = tf.reduce_mean(cross_entropy, name='cross_entropy')
# 这里的 labels 的 shape 为 [batch_size, 1]。
# 再使用 tf.add_to_collection 把 cross entropy 的 loss 添加到整体 losses 的 collection 中。
# 最后，使用 tf.add_n 将整体 losses 的 collection中 的全部 loss 求和，得到最终的 loss 并返回，
# 其中包含 cross entropy loss，还有后两个全连接层中的 weight 的 L2 loss
tf.add_to_collection('losses', cross_entropy_mean)
# The total loss is defined as the cross entropy loss plus all of the weight
# decay terms (L2 loss).
return tf.add_n(tf.get_collection('losses'), name='total_loss')
def _add_loss_summaries(total_loss):
"""Add summaries for losses in CIFAR-10 model.
Generates moving average for all losses and associated summaries for
visualizing the performance of the network.
Args:
total_loss: Total loss from loss().
Returns:
loss_averages_op: op for generating moving averages of losses.
"""
# Compute the moving average of all individual losses and the total loss.
# 创建一个新的指数滑动均值对象
loss_averages = tf.train.ExponentialMovingAverage(0.9, name='avg')
# 从字典集合中返回关键字'losses'对应的所有变量，包括交叉熵损失和正则项损失
losses = tf.get_collection('losses')
# 创建'shadow variables'并添加维护滑动均值的操作
# apply() 方***添加 trained variables 的 shadow copies，并添加操作来维护变量的滑动均值到 shadow copies。
# 滑动均值是通过指数衰减计算得到的，shadow variable 的初始化值和 trained variables 相同，
# 其更新公式为 shadow_variable = decay * shadow_variable + (1 - decay) * variable。
loss_averages_op = loss_averages.apply(losses + [total_loss])
# Attach a scalar summary to all individual losses and the total loss; do the
# same for the averaged version of the losses.
for l in losses + [total_loss]:
# Name each loss as '(raw)' and name the moving average version of the loss
# as the original loss name.
tf.summary.scalar(l.op.name + ' (raw)', l)
tf.summary.scalar(l.op.name, loss_averages.average(l))
return loss_averages_op
def train(total_loss, global_step):
"""Train CIFAR-10 model.
Create an optimizer and apply to all trainable variables. Add moving
average for all trainable variables.
Args:
total_loss: Total loss from loss().
global_step: Integer Variable counting the number of training steps
processed.
Returns:
train_op: op for training.
"""
# Variables that affect learning rate.
num_batches_per_epoch = NUM_EXAMPLES_PER_EPOCH_FOR_TRAIN / FLAGS.batch_size
decay_steps = int(num_batches_per_epoch * NUM_EPOCHS_PER_DECAY)
# Decay the learning rate exponentially based on the number of steps.
# 首先定义学习率（learning rate），并设置随迭代次数衰减，并进行 summary：
lr = tf.train.exponential_decay(INITIAL_LEARNING_RATE,
global_step,
decay_steps,
LEARNING_RATE_DECAY_FACTOR,
staircase=True)
tf.summary.scalar('learning_rate', lr)
# Generate moving averages of all losses and associated summaries.
# 对 loss 生成滑动均值和汇总，通过使用指数衰减，来维护变量的滑动均值(Moving Average)。
# 当训练模型时，维护训练参数的滑动均值是有好处的，在测试过程中使用滑动参数比最终训练的参数值本身，会提高模型的实际性能即准确率。
loss_averages_op = _add_loss_summaries(total_loss) # 损失变量的更新操作
# Compute gradients.
# 定义训练方法与目标，tf.control_dependencies 是一个 context manager，控制节点执行顺序，先执行[ ]中的操作，再执行 context 中的操作：
with tf.control_dependencies([loss_averages_op]):
opt = tf.train.GradientDescentOptimizer(lr) # 优化器 随机梯度下降法
grads = opt.compute_gradients(total_loss) # 返回计算出的(gradient, variable) pairs
# Apply gradients.
# 返回一步梯度更新操作
apply_gradient_op = opt.apply_gradients(grads, global_step=global_step)
# Add histograms for trainable variables.
for var in tf.trainable_variables():
tf.summary.histogram(var.op.name, var)
# Add histograms for gradients.
for grad, var in grads:
if grad is not None:
tf.summary.histogram(var.op.name + '/gradients', grad)
# Track the moving averages of all trainable variables.
# 最后，动态调整衰减率，返回模型参数变量的滑动更新操作即 train op：
variable_averages = tf.train.ExponentialMovingAverage(
MOVING_AVERAGE_DECAY, global_step)
variables_averages_op = variable_averages.apply(tf.trainable_variables())
with tf.control_dependencies([apply_gradient_op, variables_averages_op]):
train_op = tf.no_op(name='train')
return train_op
def maybe_download_and_extract():
"""Download and extract the tarball from Alex's website."""
dest_directory = FLAGS.data_dir
if not os.path.exists(dest_directory):
os.makedirs(dest_directory)
filename = DATA_URL.split('/')[-1]
filepath = os.path.join(dest_directory, filename)
print(filepath)
if not os.path.exists(filepath):
def _progress(count, block_size, total_size):
sys.stdout.write('\r>> Downloading %s %.1f%%' % (filename,
float(count * block_size) / float(total_size) * 100.0))
sys.stdout.flush()
filepath, _ = urllib.request.urlretrieve(DATA_URL, filepath, _progress)
print()
statinfo = os.stat(filepath)
print('Successfully downloaded', filename, statinfo.st_size, 'bytes.')
extracted_dir_path = os.path.join(dest_directory, 'cifar-10-batches-bin')
if not os.path.exists(extracted_dir_path):
tarfile.open(filepath, 'r:gz').extractall(dest_directory)
3. cifar10_train.py
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
from datetime import datetime
import time
import tensorflow as tf
import cifar10
parser = cifar10.parser
parser.add_argument('--train_dir', type=str, default='cifar10_train/',
help='Directory where to write event logs and checkpoint.')
parser.add_argument('--max_steps', type=int, default=1000000,
help='Number of batches to run.')
parser.add_argument('--log_device_placement', type=bool, default=False,
help='Whether to log device placement.')
parser.add_argument('--log_frequency', type=int, default=10,
help='How often to log results to the console.')
def train():
"""Train CIFAR-10 for a number of steps."""
# 指定当前图为默认graph
with tf.Graph().as_default():
# 设置trainable=False，是因为防止训练过程中对global_step变量也进行滑动更新操作 global_step = tf.Variable(0, trainable=False)
global_step = tf.train.get_or_create_global_step()
# Get images and labels for CIFAR-10.
# Force input pipeline to CPU:0 to avoid operations sometimes ending up on
# GPU and resulting in a slow down.
with tf.device('/cpu:0'):
images, labels = cifar10.distorted_inputs()
# Build a Graph that computes the logits predictions from the
# inference model.
logits = cifar10.inference(images)
# Calculate loss.
loss = cifar10.loss(logits, labels)
# Build a Graph that trains the model with one batch of examples and
# updates the model parameters.
train_op = cifar10.train(loss, global_step)
class _LoggerHook(tf.train.SessionRunHook):
"""Logs loss and runtime."""
def begin(self):
self._step = -1
self._start_time = time.time()
def before_run(self, run_context):
self._step += 1
return tf.train.SessionRunArgs(loss) # Asks for loss value.
def after_run(self, run_context, run_values):
if self._step % FLAGS.log_frequency == 0:
current_time = time.time()
duration = current_time - self._start_time
self._start_time = current_time
loss_value = run_values.results
examples_per_sec = FLAGS.log_frequency * FLAGS.batch_size / duration
sec_per_batch = float(duration / FLAGS.log_frequency)
format_str = ('%s: step %d, loss = %.2f (%.1f examples/sec; %.3f '
'sec/batch)')
print(format_str % (datetime.now(), self._step, loss_value,
examples_per_sec, sec_per_batch))
with tf.train.MonitoredTrainingSession(
checkpoint_dir=FLAGS.train_dir,
hooks=[tf.train.StopAtStepHook(last_step=FLAGS.max_steps),
tf.train.NanTensorHook(loss),
_LoggerHook()],
config=tf.ConfigProto(
log_device_placement=FLAGS.log_device_placement)) as mon_sess:
while not mon_sess.should_stop():
mon_sess.run(train_op)
def main(argv=None): # pylint: disable=unused-argument
cifar10.maybe_download_and_extract()
if tf.gfile.Exists(FLAGS.train_dir):
tf.gfile.DeleteRecursively(FLAGS.train_dir)
tf.gfile.MakeDirs(FLAGS.train_dir)
train()
if __name__ == '__main__':
FLAGS = parser.parse_args()
tf.app.run()
# tensorboard --logdir=D:\tmp\cifar10_train
4. cifar10_eval.py
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
from datetime import datetime
import math
import time
import numpy as np
import tensorflow as tf
import cifar10
parser = cifar10.parser
parser.add_argument('--eval_dir', type=str, default='/cifar10_eval',
help='Directory where to write event logs.')
parser.add_argument('--eval_data', type=str, default='test',
help='Either `test` or `train_eval`.')
parser.add_argument('--checkpoint_dir', type=str, default='/cifar10_train',
help='Directory where to read model checkpoints.')
parser.add_argument('--eval_interval_secs', type=int, default=60 * 5,
help='How often to run the eval.')
parser.add_argument('--num_examples', type=int, default=10000,
help='Number of examples to run.')
parser.add_argument('--run_once', type=bool, default=False,
help='Whether to run eval only once.')
# cifar10_train.py 会周期性的在检查点文件中保存模型中的所有参数，但是不会对模型进行评估。
# cifar10_eval.py 会使用该检查点文件在另一部分数据集上测试预测性能。
# 利用 inference() 函数重构模型，并使用了在评估数据集所有10,000张 CIFAR-10 图片进行测试。
# 最终计算出的精度为 1 : N，N = 预测值中置信度最高的一项与图片真实 label 匹配的频次。
# 为了监控模型在训练过程中的改进情况，评估用的脚本文件会周期性的在最新的检查点文件上运行，
# 这些检查点文件是由上述的 cifar10_train.py 产生
def eval_once(saver, summary_writer, top_k_op, summary_op):
"""Run Eval once.
Args:
saver: Saver.
summary_writer: Summary writer.
top_k_op: Top K op.
summary_op: Summary op.
"""
with tf.Session() as sess:
ckpt = tf.train.get_checkpoint_state(FLAGS.checkpoint_dir)
if ckpt and ckpt.model_checkpoint_path:
# Restores from checkpoint
saver.restore(sess, ckpt.model_checkpoint_path)
# Assuming model_checkpoint_path looks something like:
# /my-favorite-path/cifar10_train/model.ckpt-0,
# extract global_step from it.
global_step = ckpt.model_checkpoint_path.split('/')[-1].split('-')[-1]
else:
print('No checkpoint file found')
return
# Start the queue runners.
coord = tf.train.Coordinator()
try:
threads = []
for qr in tf.get_collection(tf.GraphKeys.QUEUE_RUNNERS):
threads.extend(qr.create_threads(sess, coord=coord, daemon=True,
start=True))
num_iter = int(math.ceil(FLAGS.num_examples / FLAGS.batch_size))
true_count = 0 # Counts the number of correct predictions.
total_sample_count = num_iter * FLAGS.batch_size
step = 0
while step < num_iter and not coord.should_stop():
predictions = sess.run([top_k_op])
true_count += np.sum(predictions)
step += 1
# Compute precision @ 1.
precision = true_count / total_sample_count
print('%s: precision @ 1 = %.3f' % (datetime.now(), precision))
summary = tf.Summary()
summary.ParseFromString(sess.run(summary_op))
summary.value.add(tag='Precision @ 1', simple_value=precision)
summary_writer.add_summary(summary, global_step)
except Exception as e: # pylint: disable=broad-except
coord.request_stop(e)
coord.request_stop()
coord.join(threads, stop_grace_period_secs=10)
def evaluate():
"""Eval CIFAR-10 for a number of steps."""
with tf.Graph().as_default() as g:
# Get images and labels for CIFAR-10.
eval_data = FLAGS.eval_data == 'test'
images, labels = cifar10.inputs(eval_data=eval_data)
# Build a Graph that computes the logits predictions from the
# inference model.
logits = cifar10.inference(images)
# Calculate predictions.
top_k_op = tf.nn.in_top_k(logits, labels, 1)
# Restore the moving average version of the learned variables for eval.
variable_averages = tf.train.ExponentialMovingAverage(
cifar10.MOVING_AVERAGE_DECAY)
variables_to_restore = variable_averages.variables_to_restore()
saver = tf.train.Saver(variables_to_restore)
# Build the summary operation based on the TF collection of Summaries.
summary_op = tf.summary.merge_all()
summary_writer = tf.summary.FileWriter(FLAGS.eval_dir, g)
while True:
eval_once(saver, summary_writer, top_k_op, summary_op)
if FLAGS.run_once:
break
time.sleep(FLAGS.eval_interval_secs)
def main(argv=None): # pylint: disable=unused-argument
cifar10.maybe_download_and_extract()
if tf.gfile.Exists(FLAGS.eval_dir):
tf.gfile.DeleteRecursively(FLAGS.eval_dir)
tf.gfile.MakeDirs(FLAGS.eval_dir)
evaluate()
if __name__ == '__main__':
FLAGS = parser.parse_args()
tf.app.run()
# 在训练脚本会为所有学习变量计算其滑动均值(Moving Average)，
# 评估脚本则直接将所有学习到的模型参数替换成对应的滑动均值，这一替代方式可以在评估过程中提升模型的性能。
# tensorboard --logdir=D:\tmp\cifar10_train
运行结果
cifar10_train.py的运行结果 ：
TensorBoard的可视化：
遇到的问题 && 解决方法
1. 问题一：pytharm找不到数据文件
错误信息： ValueError: Failed to find file: cifar10_data/cifar-10-batches-bin\data_batc
错误原因：
看到下面博客中的解释，发现是因为预先定义了cifar-10的存储路径，默认路径是/tmp/cifar10_train， 所以运行时找不到文件。
解决方法：
（1）将cifar10_train.py中第14行的默认路径改为了 ‘cifar10_train/’
（2）将cifar10.py中第36行的默认路径改成了cifar10数据存储的绝对路径，例：'D:/QQ文件/cifar10test1/cifar10_data/'
（然后就可以运行了）
2. 问题二：tensorboard的网址无法打开
在命令行输入tensorboard --logdir cifar10_train/可以查看训练进度， 其中 –logdir cifar10_train/ 表 示模型训练日志保存的位置：
该网址谷歌浏览器无法打开，解决方法：在原命令语句后加上" --host=127.0.0.1"，这样返回的地址也 变成完整的127.0.0.1:6006，再到浏览器将地址输入，即可正常打开网页。 重新在命令行输入：tensorboard --logdir cifar10_train/ --host=127.0.0.1
参考
https://blog.csdn.net/barry_j/article/details/79252438
https://www.cnblogs.com/gangzhucoll/p/12778246.html
https://blog.csdn.net/wang_kmin/article/details/81637816
https://www.cnblogs.com/YouXiangLiThon/articles/7246169.html
