Tensorflow样例代码分析cifar10

github地址：https://github.com/tensorflow/models.git

本文分析tutorial/image/cifar10教程项目的cifar10_input.py代码。

给外部调用的方法是：

distorted_inputs()和inputs()
cifar10.py文件调用了此文件中定义的方法。

"""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

# 定义图片的像素，原生图片32 x 32
# 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.
# IMAGE_SIZE = 24
IMAGE_SIZE = 32
# Global constants describing the CIFAR-10 data set.
# 分类数量
NUM_CLASSES = 10
# 训练集大小
NUM_EXAMPLES_PER_EPOCH_FOR_TRAIN = 50000
# 评价集大小
NUM_EXAMPLES_PER_EPOCH_FOR_EVAL = 10000

# 从CIFAR10数据文件中读取样例
# filename_queue一个队列的文件名
def read_cifar10(filename_queue):

    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.
    # 分类结果的长度，CIFAR-100长度为2
    label_bytes = 1  # 2 for CIFAR-100
    result.height = 32
    result.width = 32
    # 3位表示rgb颜色（0-255,0-255,0-255）
    result.depth = 3
    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.
    # 单个记录的总长度=分类结果长度+图片长度
    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.
    # 读取
    reader = tf.FixedLengthRecordReader(record_bytes=record_bytes)
    result.key, value = reader.read(filename_queue)

    # Convert from a string to a vector of uint8 that is record_bytes long.
    record_bytes = tf.decode_raw(value, tf.uint8)

    # 第一位代表lable-图片的正确分类结果，从uint8转换为int32类型
    # The first bytes represent the label, which we convert from uint8->int32.
    result.label = tf.cast(
        tf.strided_slice(record_bytes, [0], [label_bytes]), tf.int32)

    # 分类结果之后的数据代表图片，我们重新调整大小
    # The remaining bytes after the label represent the image, which we reshape
    # from [depth * height * width] to [depth, height, width].
    depth_major = tf.reshape(
        tf.strided_slice(record_bytes, [label_bytes],
                         [label_bytes + image_bytes]),
        [result.depth, result.height, result.width])
    # 格式转换，从[颜色,高度,宽度]--》[高度,宽度,颜色]
    # Convert from [depth, height, width] to [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):

    # Create a queue that shuffles the examples, and then
    # read 'batch_size' images + labels from the example queue.
    # 线程数
    num_preprocess_threads = 8
    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_size,
            min_after_dequeue=min_queue_examples)
    else:
        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.
    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):

    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.
    # 随机裁剪图片
    # Randomly crop a [height, width] section of the image.
    distorted_image = tf.random_crop(reshaped_image, [height, width, 3])
    # 随机旋转图片
    # 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.
    # 亮度变换
    distorted_image = tf.image.random_brightness(distorted_image,
                                                 max_delta=63)
    # 对比度变换
    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.
    # Linearly scales image to have zero mean and unit norm
    # 标准化
    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):

    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
    filename_queue = tf.train.string_input_producer(filenames)

    # Read examples from files in the filename queue.
    # 从文件中读取解析出的图片队列
    read_input = read_cifar10(filename_queue)
    # 转换为float
    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.
    # 剪切图片的中心
    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.
    # 标准化图片
    float_image = tf.image.per_image_standardization(resized_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 *
                             min_fraction_of_examples_in_queue)

    # 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=False)