读书笔记TF016葡萄娱乐场

AlexNet(亚历克斯Krizhevsky,ILSV索罗德C2013亚军)适合做图像分类。层自左向右、自上向下读取,关联层分为一组,中度、宽度减小,深度扩充。深度扩充减弱互连网总结量。

学习笔记TF016:CNN完结、数据集、TFRecord、加载图像、模型、练习、调节和测试,tf016tfrecord

AlexNet(亚历克斯Krizhevsky,ILSVEscortC二零一二亚军)适合做图像分类。层自左向右、自上向下读取,关联层分为一组,中度、宽度减小,深度增添。深度扩大减弱互联网计算量。

教练模型数据集 Stanford总计机视觉站点斯坦ford Dogs
http://vision.stanford.edu/aditya86/ImageNetDogs/
。数据下载解压到模型代码同一路径imagenet-dogs目录下。包蕴的120种狗图像。4/5教练,十分二测试。产品模型须要预留原始数据交叉验证。每幅图像JPEG格式(君越GB),尺寸不一。

图像转TFRecord文件,有助加快磨练,简化图像标签匹配,图像分离利用检查点文件对模型举行不间断测试。转换图像格式把颜色空间转灰度,图像修改统一尺寸,标签除上每幅图像。陶冶前只举行3遍预处理,时间较长。

glob.glob
枚举钦命路线目录,突显数据集文件结构。“*”通配符可以实现模糊查找。文件名中九个数字对应ImageNet系列WordNetID。ImageNet网站可用WordNetID查图像细节:
http://www.image-net.org/synset?wnid=n02085620

文本名分解为项目和呼应的文本名,品种对应文件夹名称。依照品种对图像分组。枚举每种品种图像,十分之二图像划入测试集。检查各种系列测试图像是不是至少有整套图像的18%。目录和图像组织到七个与各种项目有关的字典,包罗各档次全数图像。分类图像组织到字典中,简化接纳分类图像及分类进程。

预处理阶段,依次遍历全数分类图像,打开列表汉语件。用dataset图像填充TFRecord文件,把项目包罗进去。dataset键值对应文件列表标签。record_location
存款和储蓄TFRecord输出路径。枚举dataset,当前目录用于文书划分,每隔100m幅图像,演习样本音信写入新的TFRecord文件,加速写操作进度。不可能被TensorFlow识别为JPEG图像,用try/catch忽略。转为灰度图收缩总计量和内部存款和储蓄器占用。tf.cast把索罗德GB值转换来[0,1)区间内。标签按字符串存款和储蓄较高速,最佳转换为整数索引或独热编码秩1张量。

打开每幅图像,转换为灰度图,调整尺寸,添加到TFRecord文件。tf.image.resize_images函数把持有图像调整为同样尺寸,不考虑长度宽度比,有扭动。裁剪、边界填充能保持图像长度宽度比。

依照TFRecord文件读取图像,每一回加载少量图像及标签。修改图像形状有助练习和出口可视化。匹配全数在教练集目录下TFRecord文件加载磨炼图像。每一个TFRecord文件包括多幅图像。tf.parse_single_example只从文件提取单个样本。批运算可同时演练多幅图像或单幅图像,供给丰盛系统内部存款和储蓄器。

图像转灰度值为[0,1)浮点类型,匹配convolution2d意在输入。卷积输出第②维和最终一维不改变,中间两维产生变化。tf.contrib.layers.convolution2d创建立模型型第②层。weights_initializer设置正态随机值,第壹组滤波器填充正态分布随机数。滤波器设置trainable,消息输入网络,权值调整,升高模型准确率。
max_pool把出口降采集样品。ksize、strides
([1,2,2,1]),卷积输出形状减半。输出形状减小,不更改滤波器数量(输出通道)或图像批数量尺寸。收缩重量,与图像(滤波器)低度、宽度有关。更加多输出通道,滤波器数量增多,2倍于第①层。多个卷积和池化层减弱输入高度、宽度,扩展吃水。很多框架结构,卷积层和池化层超越5层。锻练调节和测试时间更长,能协作更加多更复杂情势。
图像各类点与出口神经元建立全连接。softmax,全连接层须要二阶张量。第壹维区分图像,第3维输入张量秩1张量。tf.reshape
提醒和行使其它全体维,-1把最后池化层调整为伟大秩1张量。
池化层展开,互连网当前场馆与猜度全连接层整合。weights_initializer接收可调用参数,lambda表达式再次来到截断正态分布,钦点分布标准差。dropout
削减模型中神经元主要性。tf.contrib.layers.fully_connected
输出后边全部层与教练中分类的全连接。各类像素与分类关联。互连网每一步将输入图像转化为滤波减小尺寸。滤波器与标签匹配。减少练习、测试网络总结量,输出更具一般性。

教练多少真实标签和模型预测结果,输入报到并且接受集磨练优化器(优化每层权值)总结模型损失。多次迭代,每一趟进步模型准确率。大多数分类函数(tf.nn.softmax)供给数值类型标签。每一个标签转换代表包括全数分类列表索引整数。tf.map_fn
匹配每一种标签并回到连串列表索引。map根据目录列表创立蕴涵分类列表。tf.map_fn
可用钦命函数对数码流图张量映射,生成仅包括每种标签在具有类标签列表索引秩1张量。tf.nn.softmax用索引预测。

调节CNN,观看滤波器(卷积核)每轮迭代变化。设计赏心悦目CNN,第3个卷积层工作,输入权值被任意初步化。权值通过图像激活,激活函数输出(特征图)随机。特征图可视化,输出外观与原始图相似,被施加静力(static)。静力由全体权值的随意激发。经过多轮迭代,权值被调整拟合练习反馈,滤波器趋于同一。互连网没有,滤波器与图像不一样细小格局类似。tf.image_summary得报到并且接受集陶冶后的滤波器和特征图简单视图。数据流图图像概要输出(image
summary
output)从全体驾驭所利用的滤波器和输入图像特点图。TensorDebugger,迭代中以GIF动画查看滤波器变化。

文件输入存储在SparseTensor,半数以上分量为0。CNN使用稠密输入,每一个值都首要,输入大多数轻重非0。

 

    import tensorflow as tf
    import glob
    from itertools import groupby
    from collections import defaultdict
    sess = tf.InteractiveSession()
    image_filenames = glob.glob("./imagenet-dogs/n02*/*.jpg")
    image_filenames[0:2]
    training_dataset = defaultdict(list)
    testing_dataset = defaultdict(list)
    image_filename_with_breed = map(lambda filename: (filename.split("/")[2], filename), image_filenames)
    for dog_breed, breed_images in groupby(image_filename_with_breed, lambda x: x[0]):
        for i, breed_image in enumerate(breed_images):
            if i % 5 == 0:
                testing_dataset[dog_breed].append(breed_image[1])
            else:
                training_dataset[dog_breed].append(breed_image[1])
        breed_training_count = len(training_dataset[dog_breed])
        breed_testing_count = len(testing_dataset[dog_breed])
        breed_training_count_float = float(breed_training_count)
        breed_testing_count_float = float(breed_testing_count)
        assert round(breed_testing_count_float / (breed_training_count_float + breed_testing_count_float), 2) > 0.18, "Not enough testing images."
    print "training_dataset testing_dataset END ------------------------------------------------------"
    def write_records_file(dataset, record_location):
        writer = None
        current_index = 0
        for breed, images_filenames in dataset.items():
            for image_filename in images_filenames:
                if current_index % 100 == 0:
                    if writer:
                        writer.close()
                    record_filename = "{record_location}-{current_index}.tfrecords".format(
                        record_location=record_location,
                        current_index=current_index)
                    writer = tf.python_io.TFRecordWriter(record_filename)
                    print record_filename + "------------------------------------------------------" 
                current_index += 1
                image_file = tf.read_file(image_filename)
                try:
                    image = tf.image.decode_jpeg(image_file)
                except:
                    print(image_filename)
                    continue
                grayscale_image = tf.image.rgb_to_grayscale(image)
                resized_image = tf.image.resize_images(grayscale_image, [250, 151])
                image_bytes = sess.run(tf.cast(resized_image, tf.uint8)).tobytes()
                image_label = breed.encode("utf-8")
                example = tf.train.Example(features=tf.train.Features(feature={
                    'label': tf.train.Feature(bytes_list=tf.train.BytesList(value=[image_label])),
                    'image': tf.train.Feature(bytes_list=tf.train.BytesList(value=[image_bytes]))
                }))
                writer.write(example.SerializeToString())
        writer.close()
    write_records_file(testing_dataset, "./output/testing-images/testing-image")
    write_records_file(training_dataset, "./output/training-images/training-image")
    print "write_records_file testing_dataset training_dataset END------------------------------------------------------"
    filename_queue = tf.train.string_input_producer(
    tf.train.match_filenames_once("./output/training-images/*.tfrecords"))
    reader = tf.TFRecordReader()
    _, serialized = reader.read(filename_queue)
    features = tf.parse_single_example(
    serialized,
        features={
            'label': tf.FixedLenFeature([], tf.string),
            'image': tf.FixedLenFeature([], tf.string),
        })
    record_image = tf.decode_raw(features['image'], tf.uint8)
    image = tf.reshape(record_image, [250, 151, 1])
    label = tf.cast(features['label'], tf.string)
    min_after_dequeue = 10
    batch_size = 3
    capacity = min_after_dequeue + 3 * batch_size
    image_batch, label_batch = tf.train.shuffle_batch(
        [image, label], batch_size=batch_size, capacity=capacity, min_after_dequeue=min_after_dequeue)
    print "load image from TFRecord END------------------------------------------------------"
    float_image_batch = tf.image.convert_image_dtype(image_batch, tf.float32)
    conv2d_layer_one = tf.contrib.layers.convolution2d(
        float_image_batch,
        num_outputs=32,
        kernel_size=(5,5),
        activation_fn=tf.nn.relu,
        weights_initializer=tf.random_normal,
        stride=(2, 2),
        trainable=True)
    pool_layer_one = tf.nn.max_pool(conv2d_layer_one,
        ksize=[1, 2, 2, 1],
        strides=[1, 2, 2, 1],
        padding='SAME')
    conv2d_layer_one.get_shape(), pool_layer_one.get_shape()
    print "conv2d_layer_one pool_layer_one END------------------------------------------------------"
    conv2d_layer_two = tf.contrib.layers.convolution2d(
        pool_layer_one,
        num_outputs=64,
        kernel_size=(5,5),
        activation_fn=tf.nn.relu,
        weights_initializer=tf.random_normal,
        stride=(1, 1),
        trainable=True)
    pool_layer_two = tf.nn.max_pool(conv2d_layer_two,
        ksize=[1, 2, 2, 1],
        strides=[1, 2, 2, 1],
        padding='SAME')
    conv2d_layer_two.get_shape(), pool_layer_two.get_shape()
    print "conv2d_layer_two pool_layer_two END------------------------------------------------------"
    flattened_layer_two = tf.reshape(
        pool_layer_two,
        [
            batch_size,
            -1
        ])
    flattened_layer_two.get_shape()
    print "flattened_layer_two END------------------------------------------------------"
    hidden_layer_three = tf.contrib.layers.fully_connected(
        flattened_layer_two,
        512,
        weights_initializer=lambda i, dtype: tf.truncated_normal([38912, 512], stddev=0.1),
        activation_fn=tf.nn.relu
    )
    hidden_layer_three = tf.nn.dropout(hidden_layer_three, 0.1)
    final_fully_connected = tf.contrib.layers.fully_connected(
        hidden_layer_three,
        120,
        weights_initializer=lambda i, dtype: tf.truncated_normal([512, 120], stddev=0.1)
    )
    print "final_fully_connected END------------------------------------------------------"
    labels = list(map(lambda c: c.split("/")[-1], glob.glob("./imagenet-dogs/*")))
    train_labels = tf.map_fn(lambda l: tf.where(tf.equal(labels, l))[0,0:1][0], label_batch, dtype=tf.int64)
    loss = tf.reduce_mean(
        tf.nn.sparse_softmax_cross_entropy_with_logits(
            final_fully_connected, train_labels))
    batch = tf.Variable(0)
    learning_rate = tf.train.exponential_decay(
        0.01,
        batch * 3,
        120,
        0.95,
        staircase=True)
    optimizer = tf.train.AdamOptimizer(
        learning_rate, 0.9).minimize(
        loss, global_step=batch)
    train_prediction = tf.nn.softmax(final_fully_connected)
    print "train_prediction END------------------------------------------------------"
    filename_queue.close(cancel_pending_enqueues=True)
    coord.request_stop()
    coord.join(threads)
    print "END------------------------------------------------------"

 

参考资料:
《面向机器智能的TensorFlow实践》

迎接加小编微信沟通:qingxingfengzi
小编的微信公众号:qingxingfengzigz
自家内人张幸清的微信公众号:qingqingfeifangz

http://www.bkjia.com/Pythonjc/1213552.htmlwww.bkjia.comtruehttp://www.bkjia.com/Pythonjc/1213552.htmlTechArticle学习笔记TF016:CNN实现、数据集、TFRecord、加载图像、模型、训练、调试,tf016tfrecord
亚历克斯Net(亚历克斯 Krizhevsky,ILSVCR-VC贰零壹叁季军)适合做图像分类。层自左…

陶冶模型数据集 Stanford总计机视觉站点Stanford Dogs
http://vision.stanford.edu/aditya86/ImageNetDogs/
。数据下载解压到模型代码同一路径imagenet-dogs目录下。包括的120种狗图像。五分四操练,伍分之一测试。产品模型供给预留原始数据交叉验证。每幅图像JPEG格式(QX56GB),尺寸不一。

图像转TFRecord文件,有助加快磨练,简化图像标签匹配,图像分离利用检查点文件对模型举行不间断测试。转换图像格式把颜色空间转灰度,图像修改统一尺寸,标签除上每幅图像。磨练前只实行三次预处理,时间较长。

glob.glob
枚举钦命路线目录,显示数据集文件结构。“*”通配符能够兑现模糊查找。文件名中几个数字对应ImageNet体系WordNetID。ImageNet网站可用WordNetID查图像细节:
http://www.image-net.org/synset?wnid=n02085620

文件名分解为品种和相应的公文名,品种对应文件夹名称。依照品种对图像分组。枚举各类门类图像,十分二图像划入测试集。检查各样项目测试图像是还是不是至少有全方位图像的18%。目录和图像组织到四个与每一种种类有关的字典,包涵各个型全体图像。分类图像组织到字典中,简化选取分类图像及分类进程。

预处理阶段,依次遍历全体分类图像,打开列表普通话件。用dataset图像填充TFRecord文件,把品种包括进去。dataset键值对应文件列表标签。record_location
存款和储蓄TFRecord输出路径。枚举dataset,当前目录用于文书划分,每隔100m幅图像,练习样本新闻写入新的TFRecord文件,加速写操作进度。一点都不大概被TensorFlow识别为JPEG图像,用try/catch忽略。转为灰度图减弱计算量和内部存款和储蓄器占用。tf.cast把XC60GB值转换成[0,1)区间内。标签按字符串存款和储蓄较高速,最棒转换为整数索引或独热编码秩1张量。

开拓每幅图像,转换为灰度图,调整尺寸,添加到TFRecord文件。tf.image.resize_images函数把富有图像调整为同样尺寸,不考虑长度宽度比,有扭动。裁剪、边界填充能保持图像长度宽度比。

规行矩步TFRecord文件读取图像,每便加载少量图像及标签。修改图像形状有助磨炼和出口可视化。匹配全体在教练集目录下TFRecord文件加载磨炼图像。每种TFRecord文件包涵多幅图像。tf.parse_single_example只从文件提取单个样本。批运算可同时演习多幅图像或单幅图像,须要丰裕系统内部存款和储蓄器。

图像转灰度值为[0,1)浮点类型,匹配convolution2d梦想输入。卷积输出第②维和最终一维不转移,中间两维爆发变化。tf.contrib.layers.convolution2d创制模型第③层。weights_initializer设置正态随机值,第二组滤波器填充正态分布随机数。滤波器设置trainable,新闻输入网络,权值调整,进步模型准确率。
max_pool把出口降采集样品。ksize、strides
([1,2,2,1]),卷积输出形状减半。输出形状减小,不改动滤波器数量(输出通道)或图像批数量尺寸。减弱重量,与图像(滤波器)中度、宽度有关。越多输出通道,滤波器数量净增,2倍于第2层。多个卷积和池化层收缩输入中度、宽度,扩大吃水。很多架构,卷积层和池化层超过5层。练习调试时间更长,能匹配更加多更复杂情势。
图像每一种点与出口神经元建立全连接。softmax,全连接层需求二阶张量。第壹维区分图像,第三维输入张量秩1张量。tf.reshape
提醒和选取其余全数维,-1把最终池化层调整为英豪秩1张量。
池化层展开,互联网当前景况与预测全连接层整合。weights_initializer接收可调用参数,lambda表明式重返截断正态分布,钦赐分布标准差。dropout
削减模型中神经元首要性。tf.contrib.layers.fully_connected
输出前面全部层与教练中分类的全连接。各类像素与分类关联。网络每一步将输入图像转化为滤波减小尺寸。滤波器与标签匹配。缩短磨练、测试互连网计算量,输出更具一般性。

教练多少真实标签和模型预测结果,输入到教练优化器(优化每层权值)计算模型损失。数十次迭代,每回提高模型准确率。抢先四分之二分拣函数(tf.nn.softmax)需要数值类型标签。每一种标签转换代表包涵全体分类列表索引整数。tf.map_fn
匹配每个标签并回到体系列表索引。map依据目录列表成立包蕴分类列表。tf.map_fn
可用钦赐函数对数码流图张量映射,生成仅包罗各类标签在装有类标签列表索引秩1张量。tf.nn.softmax用索引预测。

调剂CNN,观望滤波器(卷积核)每轮迭代变化。设计美观CNN,第②个卷积层工作,输入权值被随机初阶化。权值通过图像激活,激活函数输出(特征图)随机。特征图可视化,输出外观与原始图相似,被施加静力(static)。静力由全数权值的随机激发。经过多轮迭代,权值被调动拟合练习反馈,滤波器趋于同一。网络没有,滤波器与图像分裂细小情势类似。tf.image_summary得报到并且接受集练习后的滤波器和特色图简单视图。数据流图图像概要输出(image
summary
output)从全部精通所使用的滤波器和输入图像特点图。TensorDebugger,迭代中以GIF动画查看滤波器变化。

文本输入存款和储蓄在SparseTensor,超越二分之一分量为0。CNN使用稠密输入,各种值都主要,输入超越一半轻重非0。

 

    import tensorflow as tf
    import glob
    from itertools import groupby
    from collections import defaultdict
    sess = tf.InteractiveSession()
    image_filenames = glob.glob("./imagenet-dogs/n02*/*.jpg")
    image_filenames[0:2]
    training_dataset = defaultdict(list)
    testing_dataset = defaultdict(list)
    image_filename_with_breed = map(lambda filename: (filename.split("/")[2], filename), image_filenames)
    for dog_breed, breed_images in groupby(image_filename_with_breed, lambda x: x[0]):
        for i, breed_image in enumerate(breed_images):
            if i % 5 == 0:
                testing_dataset[dog_breed].append(breed_image[1])
            else:
                training_dataset[dog_breed].append(breed_image[1])
        breed_training_count = len(training_dataset[dog_breed])
        breed_testing_count = len(testing_dataset[dog_breed])
        breed_training_count_float = float(breed_training_count)
        breed_testing_count_float = float(breed_testing_count)
        assert round(breed_testing_count_float / (breed_training_count_float + breed_testing_count_float), 2) > 0.18, "Not enough testing images."
    print "training_dataset testing_dataset END ------------------------------------------------------"
    def write_records_file(dataset, record_location):
        writer = None
        current_index = 0
        for breed, images_filenames in dataset.items():
            for image_filename in images_filenames:
                if current_index % 100 == 0:
                    if writer:
                        writer.close()
                    record_filename = "{record_location}-{current_index}.tfrecords".format(
                        record_location=record_location,
                        current_index=current_index)
                    writer = tf.python_io.TFRecordWriter(record_filename)
                    print record_filename + "------------------------------------------------------" 
                current_index += 1
                image_file = tf.read_file(image_filename)
                try:
                    image = tf.image.decode_jpeg(image_file)
                except:
                    print(image_filename)
                    continue
                grayscale_image = tf.image.rgb_to_grayscale(image)
                resized_image = tf.image.resize_images(grayscale_image, [250, 151])
                image_bytes = sess.run(tf.cast(resized_image, tf.uint8)).tobytes()
                image_label = breed.encode("utf-8")
                example = tf.train.Example(features=tf.train.Features(feature={
                    'label': tf.train.Feature(bytes_list=tf.train.BytesList(value=[image_label])),
                    'image': tf.train.Feature(bytes_list=tf.train.BytesList(value=[image_bytes]))
                }))
                writer.write(example.SerializeToString())
        writer.close()
    write_records_file(testing_dataset, "./output/testing-images/testing-image")
    write_records_file(training_dataset, "./output/training-images/training-image")
    print "write_records_file testing_dataset training_dataset END------------------------------------------------------"
    filename_queue = tf.train.string_input_producer(
    tf.train.match_filenames_once("./output/training-images/*.tfrecords"))
    reader = tf.TFRecordReader()
    _, serialized = reader.read(filename_queue)
    features = tf.parse_single_example(
    serialized,
        features={
            'label': tf.FixedLenFeature([], tf.string),
            'image': tf.FixedLenFeature([], tf.string),
        })
    record_image = tf.decode_raw(features['image'], tf.uint8)
    image = tf.reshape(record_image, [250, 151, 1])
    label = tf.cast(features['label'], tf.string)
    min_after_dequeue = 10
    batch_size = 3
    capacity = min_after_dequeue + 3 * batch_size
    image_batch, label_batch = tf.train.shuffle_batch(
        [image, label], batch_size=batch_size, capacity=capacity, min_after_dequeue=min_after_dequeue)
    print "load image from TFRecord END------------------------------------------------------"
    float_image_batch = tf.image.convert_image_dtype(image_batch, tf.float32)
    conv2d_layer_one = tf.contrib.layers.convolution2d(
        float_image_batch,
        num_outputs=32,
        kernel_size=(5,5),
        activation_fn=tf.nn.relu,
        weights_initializer=tf.random_normal,
        stride=(2, 2),
        trainable=True)
    pool_layer_one = tf.nn.max_pool(conv2d_layer_one,
        ksize=[1, 2, 2, 1],
        strides=[1, 2, 2, 1],
        padding='SAME')
    conv2d_layer_one.get_shape(), pool_layer_one.get_shape()
    print "conv2d_layer_one pool_layer_one END------------------------------------------------------"
    conv2d_layer_two = tf.contrib.layers.convolution2d(
        pool_layer_one,
        num_outputs=64,
        kernel_size=(5,5),
        activation_fn=tf.nn.relu,
        weights_initializer=tf.random_normal,
        stride=(1, 1),
        trainable=True)
    pool_layer_two = tf.nn.max_pool(conv2d_layer_two,
        ksize=[1, 2, 2, 1],
        strides=[1, 2, 2, 1],
        padding='SAME')
    conv2d_layer_two.get_shape(), pool_layer_two.get_shape()
    print "conv2d_layer_two pool_layer_two END------------------------------------------------------"
    flattened_layer_two = tf.reshape(
        pool_layer_two,
        [
            batch_size,
            -1
        ])
    flattened_layer_two.get_shape()
    print "flattened_layer_two END------------------------------------------------------"
    hidden_layer_three = tf.contrib.layers.fully_connected(
        flattened_layer_two,
        512,
        weights_initializer=lambda i, dtype: tf.truncated_normal([38912, 512], stddev=0.1),
        activation_fn=tf.nn.relu
    )
    hidden_layer_three = tf.nn.dropout(hidden_layer_three, 0.1)
    final_fully_connected = tf.contrib.layers.fully_connected(
        hidden_layer_three,
        120,
        weights_initializer=lambda i, dtype: tf.truncated_normal([512, 120], stddev=0.1)
    )
    print "final_fully_connected END------------------------------------------------------"
    labels = list(map(lambda c: c.split("/")[-1], glob.glob("./imagenet-dogs/*")))
    train_labels = tf.map_fn(lambda l: tf.where(tf.equal(labels, l))[0,0:1][0], label_batch, dtype=tf.int64)
    loss = tf.reduce_mean(
        tf.nn.sparse_softmax_cross_entropy_with_logits(
            final_fully_connected, train_labels))
    batch = tf.Variable(0)
    learning_rate = tf.train.exponential_decay(
        0.01,
        batch * 3,
        120,
        0.95,
        staircase=True)
    optimizer = tf.train.AdamOptimizer(
        learning_rate, 0.9).minimize(
        loss, global_step=batch)
    train_prediction = tf.nn.softmax(final_fully_connected)
    print "train_prediction END------------------------------------------------------"
    filename_queue.close(cancel_pending_enqueues=True)
    coord.request_stop()
    coord.join(threads)
    print "END------------------------------------------------------"

 

参考资料:
《面向机器智能的TensorFlow实践》

欢迎加我微信交换:qingxingfengzi
本身的微信公众号:qingxingfengzigz
本人老婆张幸清的微信公众号:qingqingfeifangz