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)

Tensorflow样例代码分析cifar10的更多相关文章

  1. TensorFlow入门之MNIST样例代码分析

    这几天想系统的学习一下TensorFlow,为之后的工作打下一些基础.看了下<TensorFlow:实战Google深度学习框架>这本书,目前个人觉得这本书还是对初学者挺友好的,作者站在初 ...

  2. amaze样例页面分析(一)

    amaze样例页面分析(一) 一.总结 1.从审查(inspect)中是很清楚的可以弄清楚这些part之间的结构关系的 2.一者在于弄清楚他们之间的结构关系,二者在于知道结构的每一部分是干嘛的 3.i ...

  3. 使用ffmpeg实现转码样例(代码实现)

    分类: C/C++ 使用ffmpeg实现转码样例(代码实现) 使用ffmpeg转码主要工作如下: Demux -> Decoding -> Encoding -> Muxing 其中 ...

  4. java 线程、线程池基本应用演示样例代码回想

    java 线程.线程池基本应用演示样例代码回想 package org.rui.thread; /** * 定义任务 * * @author lenovo * */ public class Lift ...

  5. ECharts组件应用样例代码

    一.从Echarts官网上下载最新版本组件 Echarts是百度开发的开源Web图表组件,界面美观,使用简单.组件下载地址:http://echarts.baidu.com/echarts2/doc/ ...

  6. java文件夹相关操作 演示样例代码

    java文件夹相关操作 演示样例代码 package org.rui.io; import java.io.File; import java.io.FilenameFilter; import ja ...

  7. 10分钟理解Android数据库的创建与使用(附具体解释和演示样例代码)

    1.Android数据库简单介绍. Android系统的framework层集成了Sqlite3数据库.我们知道Sqlite3是一种轻量级的高效存储的数据库. Sqlite数据库具有以下长处: (1) ...

  8. C#调用 Oracle 存储过程样例代码

    -- 建表 CREATE TABLE sale_report (      sale_date DATE NOT NULL ,      sale_item VARCHAR(2) NOT NULL , ...

  9. java 又一次抛出异常 相关处理结果演示样例代码

    java 又一次抛出异常 相关处理结果演示样例代码 package org.rui.ExceptionTest; /** * 又一次抛出异常 * 在某些情况下,我们想又一次掷出刚才产生过的违例,特别是 ...

随机推荐

  1. APUE(8)---进程控制(1)

    一.进程标识 每个进程都有一个非负整型标识的唯一进程ID.因为进程ID标识符总是唯一的,常将其用做其他标识符的一部分以保证其唯一性.进程ID虽然是唯一的, 但是却是可以复用的.ID为0的进程通常是调度 ...

  2. JSP Servlet中Request与Response所有成员方法的研究

    HttpServletRequest与HttpServletResponse作为Servlet中doGet.doPost等方法中传递的参数,承接了Http请求与响应中的大部分功能,请求的解析与响应的返 ...

  3. shiro中移除jsessionid的解决方案

    在web.xml配置文件中设置 <session-config> <!-- Disables URL-based sessions (no more 'jsessionid' in ...

  4. vmware获取主机、数据中心等对象ManagedObjectReference

    在vmware的api中提供以下列表中的对象,称作ManagedObjectReference,包括虚拟机信息.主机.数据中心等等一些信息,我们可以通过vcenter的web api得到. 下面我们来 ...

  5. Question | 网站被黑客扫描撞库该怎么应对防范?

    本文来自网易云社区 在安全领域向来是先知道如何攻,其次才是防.针对题主的问题,在介绍如何防范网站被黑客扫描撞库之前,先简单介绍一下什么是撞库. 撞库是黑客通过收集互联网已泄露的用户和密码信息,生成对于 ...

  6. python-数值类型转换

    常用的数据类型转换 函数 说明 int(x [,base ]) 将x转换为一个整数 long(x [,base ]) 将x转换为一个长整数 float(x ) 将x转换到一个浮点数 complex(r ...

  7. 趣图:当我捕获Bug的时候

      趣图:当我以为已捕获了所有可能的异常...的时候 趣图:程序员调 Bug 的感觉,就是这样的

  8. MySQL开启日志记录查询/执行过的SQL语句

    作为后端开发者,遇到数据库问题的时候应该通过分析SQL语句来跟进问题所在,该方法可以记录所有的查询/执行的SQL语句到日志文件. 方法有几种,但是个人觉得以下这种最简单,但是重启MySQL服务后需要重 ...

  9. [ActionScript 3.0] 实现放大镜效果的简单方法

    //mc和bgmc是同一对象的不同实例 //mc放大的对象 //bgmc源对象 //mag放大镜 var scale:Number = 1.3;//放大倍数 mc.mask = mag; mag.st ...

  10. The server of Nginx(二)——Nginx基本功能配置

    一.Nginx访问控制 (1)基于授权的访问控制 Nginx于Apache一样,可以实现基于用户授权的访问控制,当客户端要访问相应网站或者目录时要求输入用户名密码才能正常访问,配置步骤与Apache基 ...