TensorFlow使用inception模型进行flower识别训练+修改原始inception实现predict

这是我对原始Inception做了修改之后传到github上的github链接

主要的修改是实现了输出filename和对应的label 并提供了运行相关py文件的运行命令

有需要的可以去下载

自己尝试运行inception 中的flowers_train.py 在此过程中 遇到了一些坑 在网上也没有找到详细讲述如何运行的博客

在这里写下来与大家分享我作为一个小白遇到的问题和解决的办法

TensorFlow提供了很多模型代码models 下载到本地解压  只使用其中的inception

用PyCharm打开 运行flowers_train.py（可能需要修改编译环境 选择tensorflow）

此时报错 大概就是提示inception.XXX 用不了

此时需要在第二层inception文件夹中新建空的py文件 “__init__.py”

同样的还会提示slim.XXX 用不了 在slim文件夹中新建空的py文件 “__init__.py”

再运行flowers_train.py 还是无法运行  提示找不到数据

因为我用的mac 于是就选择使用inception/inception/data/download_and_preprocess_flowers_mac.sh

将这个.sh文件拖到命令行窗口中 后面需要加上保存文件的路径 然后执行

执行过程中 一开始是没问题的 下载flower数据（此过程中最好打开 vpn）

但是下载完毕之后呢 根据.sh命令 需要将数据分为 train和validation 

然后我这就报错了 大概的意思就是gshuf commound无法执行

这个解决方案百度吧 就是安装一个东西就可以了。。。

同时呢 还有个报错 就是找不到build_image_data

这个呢 我一开始以为是需要添加一个build_image_data文件夹或者文件 发现不对

后来发现有一个build_image_data.py的文件 就按照路径要求 把这个文件复制过去了 发现还是有点问题

于是呢 看到了这个解决方法 于是就把里面的shell文件内容 复制过来 执行 得到一个输出

输出的内容有关于build_image_data.py的执行命令 于是就在terminal中 进入到build_image_data.py

所在的路径 运行指令 就可以将下载的图片转换为TFRecord 

以下我修改后的shell文件内容

#!/bin/bash# Copyright 2016 Google Inc. All Rights Reserved.## Licensed under the Apache License, Version 2.0 (the "License");# you may not use this file except in compliance with the License.# You may obtain a copy of the License at##     http://www.apache.org/licenses/LICENSE-2.0## Unless required by applicable law or agreed to in writing, software# distributed under the License is distributed on an "AS IS" BASIS,# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.# See the License for the specific language governing permissions and# limitations under the License.# ==============================================================================# Script to download and preprocess the flowers data set. This data set# provides a demonstration for how to perform fine-tuning (i.e. tranfer# learning) from one model to a new data set.## This script provides a demonstration for how to prepare an arbitrary# data set for training an Inception v3 model.## We demonstrate this with the flowers data set which consists of images# of labeled flower images from 5 classes:## daisy, dandelion, roses, sunflowers, tulips## The final output of this script are sharded TFRecord files containing# serialized Example protocol buffers. See build_image_data.py for# details of how the Example protocol buffer contains image data.## usage:#  ./download_and_preprocess_flowers.sh [data-dir]set -eif [ -z "$1" ]; then  echo "Usage: download_and_preprocess_flowers.sh [data dir]"  exitfi# Create the output and temporary directories.DATA_DIR="${1%/}"SCRATCH_DIR="${DATA_DIR}/raw-data/"mkdir -p "${DATA_DIR}"mkdir -p "${SCRATCH_DIR}"WORK_DIR="$0.runfiles/inception/inception"# Download the flowers data.DATA_URL="http://download.tensorflow.org/example_images/flower_photos.tgz"CURRENT_DIR=$(pwd)cd "${DATA_DIR}"TARBALL="flower_photos.tgz"if [ ! -f ${TARBALL} ]; then  echo "Downloading flower data set."  curl -o ${TARBALL} "${DATA_URL}"else  echo "Skipping download of flower data."fi#echo ${WORK_DIR}#/Users/youngkl/Desktop/inception/inception/tmp#echo ${DATA_DIR}#/Users/youngkl/Desktop/inception/inception/tmp# Note the locations of the train and validation data.TRAIN_DIRECTORY="${SCRATCH_DIR}train/"VALIDATION_DIRECTORY="${SCRATCH_DIR}validation/"# Expands the data into the flower_photos/ directory and rename it as the# train directory.tar xf flower_photos.tgzrm -rf "${TRAIN_DIRECTORY}" "${VALIDATION_DIRECTORY}"mv flower_photos "${TRAIN_DIRECTORY}"# Generate a list of 5 labels: daisy, dandelion, roses, sunflowers, tulipsLABELS_FILE="${SCRATCH_DIR}/labels.txt"ls -1 "${TRAIN_DIRECTORY}" | grep -v 'LICENSE' | sed 's/\///' | sort > "${LABELS_FILE}"# Generate the validation data set.while read LABEL; do  VALIDATION_DIR_FOR_LABEL="${VALIDATION_DIRECTORY}${LABEL}"  TRAIN_DIR_FOR_LABEL="${TRAIN_DIRECTORY}${LABEL}"  # Move the first randomly selected 100 images to the validation set.  mkdir -p "${VALIDATION_DIR_FOR_LABEL}"  VALIDATION_IMAGES=$(ls -1 "${TRAIN_DIR_FOR_LABEL}" | gshuf | head -100)  for IMAGE in ${VALIDATION_IMAGES}; do    mv -f "${TRAIN_DIRECTORY}${LABEL}/${IMAGE}" "${VALIDATION_DIR_FOR_LABEL}"  donedone < "${LABELS_FILE}"# Build the TFRecords version of the image data.cd "${CURRENT_DIR}"BUILD_SCRIPT="${WORK_DIR}/build_image_data"OUTPUT_DIRECTORY="${DATA_DIR}"echo "${BUILD_SCRIPT}"echo "${CURRENT_DIR}"echo "python build_image_data.py  --train_directory=${TRAIN_DIRECTORY}  --validation_directory=${VALIDATION_DIRECTORY}  --output_directory=${OUTPUT_DIRECTORY} --labels_file=${LABELS_FILE}"

数据有了之后 继续运行flowers_train.py 发现还是有问题  大概还是找不到数据的提示信息

大概想到 这个py文件执行是需要输入参数的 

查看点击打开链接 中间部分   How to Retrain a Trained Model on the Flowers Data

在这之下 发现有与之前出现的.sh文件有类似的部分 想到同样的办法 把参数具体内容输出

于是我 新建了一个flower.sh文件  文件内容如下

cd "${CURRENT_DIR}"# Directory where the flowers data resides.FLOWERS_DATA_DIR=/Users/youngkl/Desktop/inception/inception/tmp/raw-data/# Directory where to save the checkpoint and events files.TRAIN_DIR=/Users/youngkl/Desktop/inception/inception/tmp/echo "python flowers_train.py  --train_directory=${TRAIN_DIR}   --data_dir=${FLOWERS_DATA_DIR}   --fine_tune=False   --initial_learning_rate=0.001   --input_queue_memory_factor=1"#python flowers_train.py  --train_directory=/Users/youngkl/Desktop/inception/inception/tmp/   --data_dir=/Users/youngkl/Desktop/inception/inception/tmp/raw-data/   --fine_tune=False   --initial_learning_rate=0.001   --input_queue_memory_factor=1

注意 需要对其中的DIR值做修改 将这个.sh文件拖到terminal 回车执行 输出了想要的内容

复制 在terminal中 进入到flowers_train.py所在的文件目录 执行就可以了

python flowers_train.py  --train_directory=/Users/youngkl/Desktop/inception/inception/tmp/   --data_dir=/Users/youngkl/Desktop/inception/inception/tmp/raw-data/   --fine_tune=False   --initial_learning_rate=0.001   --input_queue_memory_factor=1

之后发现在这个命令中还可以添加其他的内容 比如gpu数量 最多迭代的次数 还可以设置使用预先训练好的模型进一步调节  

具体的都可以在inception_train.py中进行查看

以上是训练阶段遇到的问题 后来在测试阶段又有些问题了

一开始validation的时候呢 是没有问题的 可以输出top1 precision和top5 recall 问题出现在test阶段

在test阶段 我想输出文件名对应的label

但是原始的inception_eval.py的_eval_once函数里 并没有输出 所以我们需要进行修改

首先呢 原始的evaluate函数里面 得到了labels和logits值 将这个值传到_eval_once进行输出即可

注意 tensorflow中  logits值需要用tf.nn.softmax 才得到网络对每个类别预测的概率  概率最大的id就是预测的类别（数组中id是从0开始算的）

但是预测时输入很多图片 没办法得到输出的类别对应的哪个文件 此时需要输出文件名 但是原始的程序里面是没有的 所以需要做修改

这里给出修改的代码

inception_eval.py

# Copyright 2016 Google Inc. All Rights Reserved.## Licensed under the Apache License, Version 2.0 (the "License");# you may not use this file except in compliance with the License.# You may obtain a copy of the License at##     http://www.apache.org/licenses/LICENSE-2.0## Unless required by applicable law or agreed to in writing, software# distributed under the License is distributed on an "AS IS" BASIS,# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.# See the License for the specific language governing permissions and# limitations under the License.# =============================================================================="""A library to evaluate Inception on a single GPU."""from __future__ import absolute_importfrom __future__ import divisionfrom __future__ import print_functionfrom datetime import datetimeimport mathimport os.pathimport timeimport numpy as npimport tensorflow as tffrom inception import image_processingfrom inception import inception_model as inceptionFLAGS = tf.app.flags.FLAGStf.app.flags.DEFINE_string('eval_dir', '/tmp/imagenet_eval',                           """Directory where to write event logs.""")tf.app.flags.DEFINE_string('checkpoint_dir', '/home/yangkunlin/home/',                           """Directory where to read model checkpoints.""")# Flags governing the frequency of the eval.tf.app.flags.DEFINE_integer('eval_interval_secs', 60 * 5,                            """How often to run the eval.""")tf.app.flags.DEFINE_boolean('run_once', False,                            """Whether to run eval only once.""")# Flags governing the data used for the eval.tf.app.flags.DEFINE_integer('num_examples', 2813,                            """Number of examples to run. Note that the eval """                            """ImageNet dataset contains 50000 examples.""")tf.app.flags.DEFINE_string('subset', 'validation',                           """Either 'validation' or 'train'.""")def _eval_once(saver, summary_writer, filenames, logits, labels, top_1_op, top_5_op, summary_op):  """Runs Eval once.  Args:    saver: Saver.    summary_writer: Summary writer.    top_1_op: Top 1 op.    top_5_op: Top 5 op.    summary_op: Summary op.  """  print ("path")  print (FLAGS.checkpoint_dir)  with tf.Session() as sess:    ckpt = tf.train.get_checkpoint_state(FLAGS.checkpoint_dir)    if ckpt and ckpt.model_checkpoint_path:      if os.path.isabs(ckpt.model_checkpoint_path):        # Restores from checkpoint with absolute path.        saver.restore(sess, ckpt.model_checkpoint_path)      else:        # Restores from checkpoint with relative path.        saver.restore(sess, os.path.join(FLAGS.checkpoint_dir,                                         ckpt.model_checkpoint_path))      # Assuming model_checkpoint_path looks something like:      #   /my-favorite-path/imagenet_train/model.ckpt-0,      # extract global_step from it.      global_step = ckpt.model_checkpoint_path.split('/')[-1].split('-')[-1]      print('Successfully loaded model from %s at step=%s.' %            (ckpt.model_checkpoint_path, global_step))    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))      # Counts the number of correct predictions.      count_top_1 = 0.0      count_top_5 = 0.0      total_sample_count = num_iter * FLAGS.batch_size      step = 0      print('%s: starting evaluation on (%s).' % (datetime.now(), FLAGS.subset))      start_time = time.time()      while step < num_iter and not coord.should_stop():        filenames_,logits_, labels_, top_1, top_5 = sess.run([filenames,logits, labels, top_1_op, top_5_op])        # print (tf.nn.softmax(logits_).eval())        print (filenames_)        logi = tf.nn.softmax(logits_).eval()        # print (labels_)        # print (logits_.shape[0])        # print (logits_.shape[1])        row = logits_.shape[0]        col = logits_.shape[1]        for i in range(row):          # print (filenames[i].eval())          x = -1.0          id = -1          for j in range(col):            if logi[i][j] > x:              x = logi[i][j]              id = j          print (id)        count_top_1 += np.sum(top_1)        count_top_5 += np.sum(top_5)        step += 1        if step % 20 == 0:          duration = time.time() - start_time          sec_per_batch = duration / 20.0          examples_per_sec = FLAGS.batch_size / sec_per_batch          print('%s: [%d batches out of %d] (%.1f examples/sec; %.3f'                'sec/batch)' % (datetime.now(), step, num_iter,                                examples_per_sec, sec_per_batch))          start_time = time.time()      # Compute precision @ 1.      precision_at_1 = count_top_1 / total_sample_count      recall_at_5 = count_top_5 / total_sample_count      print('%s: precision @ 1 = %.4f recall @ 5 = %.4f [%d examples]' %            (datetime.now(), precision_at_1, recall_at_5, total_sample_count))      summary = tf.Summary()      summary.ParseFromString(sess.run(summary_op))      summary.value.add(tag='Precision @ 1', simple_value=precision_at_1)      summary.value.add(tag='Recall @ 5', simple_value=recall_at_5)      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)    # evaluate(FLAGS.subset)def evaluate(dataset):  """Evaluate model on Dataset for a number of steps."""  with tf.Graph().as_default():    # Get images and labels from the dataset.    # with tf.Session() as sess:    #   images, labels = image_processing.inputs(dataset)    #   print ("lable")    #   sess.run (labels.eval())    images, labels, filenames = image_processing.inputs(dataset)    # Number of classes in the Dataset label set plus 1.    # Label 0 is reserved for an (unused) background class.    num_classes = dataset.num_classes() + 1    # Build a Graph that computes the logits predictions from the    # inference model.    logits, _ = inception.inference(images, num_classes)    print ("logits")    # print (tf.cast(logits,tf.float32).eval())    # print ("_")    # print (tf.cast(_,tf.float32).eval())    # Calculate predictions.    top_1_op = tf.nn.in_top_k(logits, labels, 1)    top_5_op = tf.nn.in_top_k(logits, labels, 5)    # Restore the moving average version of the learned variables for eval.    variable_averages = tf.train.ExponentialMovingAverage(        inception.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()    graph_def = tf.get_default_graph().as_graph_def()    summary_writer = tf.summary.FileWriter(FLAGS.eval_dir,                                            graph_def=graph_def)    while True:      _eval_once(saver, summary_writer, filenames, logits, labels, top_1_op, top_5_op, summary_op)      if FLAGS.run_once:        break      time.sleep(FLAGS.eval_interval_secs)        # sess = tf.InteractiveSession()    # print("label")    # # label_ = sess.run([labels])    # print(labels.eval())    # sess.close()

image_procession.py

# Copyright 2016 Google Inc. All Rights Reserved.## Licensed under the Apache License, Version 2.0 (the "License");# you may not use this file except in compliance with the License.# You may obtain a copy of the License at##     http://www.apache.org/licenses/LICENSE-2.0## Unless required by applicable law or agreed to in writing, software# distributed under the License is distributed on an "AS IS" BASIS,# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.# See the License for the specific language governing permissions and# limitations under the License.# =============================================================================="""Read and preprocess image data. Image processing occurs on a single image at a time. Image are read and preprocessed in parallel across multiple threads. The resulting images are concatenated together to form a single batch for training or evaluation. -- Provide processed image data for a network: inputs: Construct batches of evaluation examples of images. distorted_inputs: Construct batches of training examples of images. batch_inputs: Construct batches of training or evaluation examples of images. -- Data processing: parse_example_proto: Parses an Example proto containing a training example   of an image. -- Image decoding: decode_jpeg: Decode a JPEG encoded string into a 3-D float32 Tensor. -- Image preprocessing: image_preprocessing: Decode and preprocess one image for evaluation or training distort_image: Distort one image for training a network. eval_image: Prepare one image for evaluation. distort_color: Distort the color in one image for training."""from __future__ import absolute_importfrom __future__ import divisionfrom __future__ import print_functionimport tensorflow as tfFLAGS = tf.app.flags.FLAGStf.app.flags.DEFINE_integer('batch_size', 32,                            """Number of images to process in a batch.""")tf.app.flags.DEFINE_integer('image_size', 299,                            """Provide square images of this size.""")tf.app.flags.DEFINE_integer('num_preprocess_threads', 4,                            """Number of preprocessing threads per tower. """                            """Please make this a multiple of 4.""")tf.app.flags.DEFINE_integer('num_readers', 4,                            """Number of parallel readers during train.""")# Images are preprocessed asynchronously using multiple threads specified by# --num_preprocss_threads and the resulting processed images are stored in a# random shuffling queue. The shuffling queue dequeues --batch_size images# for processing on a given Inception tower. A larger shuffling queue guarantees# better mixing across examples within a batch and results in slightly higher# predictive performance in a trained model. Empirically,# --input_queue_memory_factor=16 works well. A value of 16 implies a queue size# of 1024*16 images. Assuming RGB 299x299 images, this implies a queue size of# 16GB. If the machine is memory limited, then decrease this factor to# decrease the CPU memory footprint, accordingly.tf.app.flags.DEFINE_integer('input_queue_memory_factor', 16,                            """Size of the queue of preprocessed images. """                            """Default is ideal but try smaller values, e.g. """                            """4, 2 or 1, if host memory is constrained. See """                            """comments in code for more details.""")# def inputs(dataset, batch_size=None, num_preprocess_threads=None):#   """Generate batches of ImageNet images for evaluation.##   Use this function as the inputs for evaluating a network.##   Note that some (minimal) image preprocessing occurs during evaluation#   including central cropping and resizing of the image to fit the network.##   Args:#     dataset: instance of Dataset class specifying the dataset.#     batch_size: integer, number of examples in batch#     num_preprocess_threads: integer, total number of preprocessing threads but#       None defaults to FLAGS.num_preprocess_threads.##   Returns:#     images: Images. 4D tensor of size [batch_size, FLAGS.image_size,#                                        image_size, 3].#     labels: 1-D integer Tensor of [FLAGS.batch_size].#   """#   if not batch_size:#     batch_size = FLAGS.batch_size##   # Force all input processing onto CPU in order to reserve the GPU for#   # the forward inference and back-propagation.#   with tf.device('/cpu:0'):#     images, labels = batch_inputs(#         dataset, batch_size, train=False,#         num_preprocess_threads=num_preprocess_threads,#         num_readers=1)##   return images, labels### def distorted_inputs(dataset, batch_size=None, num_preprocess_threads=None):#   """Generate batches of distorted versions of ImageNet images.##   Use this function as the inputs for training a network.##   Distorting images provides a useful technique for augmenting the data#   set during training in order to make the network invariant to aspects#   of the image that do not effect the label.##   Args:#     dataset: instance of Dataset class specifying the dataset.#     batch_size: integer, number of examples in batch#     num_preprocess_threads: integer, total number of preprocessing threads but#       None defaults to FLAGS.num_preprocess_threads.##   Returns:#     images: Images. 4D tensor of size [batch_size, FLAGS.image_size,#                                        FLAGS.image_size, 3].#     labels: 1-D integer Tensor of [batch_size].#   """#   if not batch_size:#     batch_size = FLAGS.batch_size##   # Force all input processing onto CPU in order to reserve the GPU for#   # the forward inference and back-propagation.#   with tf.device('/cpu:0'):#     images, labels = batch_inputs(#         dataset, batch_size, train=True,#         num_preprocess_threads=num_preprocess_threads,#         num_readers=FLAGS.num_readers)#   return images, labels### def decode_jpeg(image_buffer, scope=None):#   """Decode a JPEG string into one 3-D float image Tensor.##   Args:#     image_buffer: scalar string Tensor.#     scope: Optional scope for name_scope.#   Returns:#     3-D float Tensor with values ranging from [0, 1).#   """#   with tf.name_scope(values=[image_buffer], name=scope,#                      default_name='decode_jpeg'):#     # Decode the string as an RGB JPEG.#     # Note that the resulting image contains an unknown height and width#     # that is set dynamically by decode_jpeg. In other words, the height#     # and width of image is unknown at compile-time.#     image = tf.image.decode_jpeg(image_buffer, channels=3)##     # After this point, all image pixels reside in [0,1)#     # until the very end, when they're rescaled to (-1, 1).  The various#     # adjust_* ops all require this range for dtype float.#     image = tf.image.convert_image_dtype(image, dtype=tf.float32)#     return image### def distort_color(image, thread_id=0, scope=None):#   """Distort the color of the image.##   Each color distortion is non-commutative and thus ordering of the color ops#   matters. Ideally we would randomly permute the ordering of the color ops.#   Rather then adding that level of complication, we select a distinct ordering#   of color ops for each preprocessing thread.##   Args:#     image: Tensor containing single image.#     thread_id: preprocessing thread ID.#     scope: Optional scope for name_scope.#   Returns:#     color-distorted image#   """#   with tf.name_scope(values=[image], name=scope, default_name='distort_color'):#     color_ordering = thread_id % 2##     if color_ordering == 0:#       image = tf.image.random_brightness(image, max_delta=32. / 255.)#       image = tf.image.random_saturation(image, lower=0.5, upper=1.5)#       image = tf.image.random_hue(image, max_delta=0.2)#       image = tf.image.random_contrast(image, lower=0.5, upper=1.5)#     elif color_ordering == 1:#       image = tf.image.random_brightness(image, max_delta=32. / 255.)#       image = tf.image.random_contrast(image, lower=0.5, upper=1.5)#       image = tf.image.random_saturation(image, lower=0.5, upper=1.5)#       image = tf.image.random_hue(image, max_delta=0.2)##     # The random_* ops do not necessarily clamp.#     image = tf.clip_by_value(image, 0.0, 1.0)#     return image### def distort_image(image, height, width, bbox, thread_id=0, scope=None):#   """Distort one image for training a network.##   Distorting images provides a useful technique for augmenting the data#   set during training in order to make the network invariant to aspects#   of the image that do not effect the label.##   Args:#     image: 3-D float Tensor of image#     height: integer#     width: integer#     bbox: 3-D float Tensor of bounding boxes arranged [1, num_boxes, coords]#       where each coordinate is [0, 1) and the coordinates are arranged#       as [ymin, xmin, ymax, xmax].#     thread_id: integer indicating the preprocessing thread.#     scope: Optional scope for name_scope.#   Returns:#     3-D float Tensor of distorted image used for training.#   """#   with tf.name_scope(values=[image, height, width, bbox], name=scope,#                      default_name='distort_image'):#     # Each bounding box has shape [1, num_boxes, box coords] and#     # the coordinates are ordered [ymin, xmin, ymax, xmax].##     # Display the bounding box in the first thread only.#     if not thread_id:#       image_with_box = tf.image.draw_bounding_boxes(tf.expand_dims(image, 0),#                                                     bbox)#       tf.summary.image('image_with_bounding_boxes', image_with_box)##   # A large fraction of image datasets contain a human-annotated bounding#   # box delineating the region of the image containing the object of interest.#   # We choose to create a new bounding box for the object which is a randomly#   # distorted version of the human-annotated bounding box that obeys an allowed#   # range of aspect ratios, sizes and overlap with the human-annotated#   # bounding box. If no box is supplied, then we assume the bounding box is#   # the entire image.#     sample_distorted_bounding_box = tf.image.sample_distorted_bounding_box(#         tf.shape(image),#         bounding_boxes=bbox,#         min_object_covered=0.1,#         aspect_ratio_range=[0.75, 1.33],#         area_range=[0.05, 1.0],#         max_attempts=100,#         use_image_if_no_bounding_boxes=True)#     bbox_begin, bbox_size, distort_bbox = sample_distorted_bounding_box#     if not thread_id:#       image_with_distorted_box = tf.image.draw_bounding_boxes(#           tf.expand_dims(image, 0), distort_bbox)#       tf.summary.image('images_with_distorted_bounding_box',#                        image_with_distorted_box)##     # Crop the image to the specified bounding box.#     distorted_image = tf.slice(image, bbox_begin, bbox_size)##     # This resizing operation may distort the images because the aspect#     # ratio is not respected. We select a resize method in a round robin#     # fashion based on the thread number.#     # Note that ResizeMethod contains 4 enumerated resizing methods.#     resize_method = thread_id % 4#     distorted_image = tf.image.resize_images(distorted_image, [height, width],#                                              method=resize_method)#     # Restore the shape since the dynamic slice based upon the bbox_size loses#     # the third dimension.#     distorted_image.set_shape([height, width, 3])#     if not thread_id:#       tf.summary.image('cropped_resized_image',#                        tf.expand_dims(distorted_image, 0))##     # Randomly flip the image horizontally.#     distorted_image = tf.image.random_flip_left_right(distorted_image)##     # Randomly distort the colors.#     distorted_image = distort_color(distorted_image, thread_id)##     if not thread_id:#       tf.summary.image('final_distorted_image',#                        tf.expand_dims(distorted_image, 0))#     return distorted_image### def eval_image(image, height, width, scope=None):#   """Prepare one image for evaluation.##   Args:#     image: 3-D float Tensor#     height: integer#     width: integer#     scope: Optional scope for name_scope.#   Returns:#     3-D float Tensor of prepared image.#   """#   with tf.name_scope(values=[image, height, width], name=scope,#                      default_name='eval_image'):#     # Crop the central region of the image with an area containing 87.5% of#     # the original image.#     image = tf.image.central_crop(image, central_fraction=0.875)##     # Resize the image to the original height and width.#     image = tf.expand_dims(image, 0)#     image = tf.image.resize_bilinear(image, [height, width],#                                      align_corners=False)#     image = tf.squeeze(image, [0])#     return image### def image_preprocessing(image_buffer, bbox, train, thread_id=0):#   """Decode and preprocess one image for evaluation or training.##   Args:#     image_buffer: JPEG encoded string Tensor#     bbox: 3-D float Tensor of bounding boxes arranged [1, num_boxes, coords]#       where each coordinate is [0, 1) and the coordinates are arranged as#       [ymin, xmin, ymax, xmax].#     train: boolean#     thread_id: integer indicating preprocessing thread##   Returns:#     3-D float Tensor containing an appropriately scaled image##   Raises:#     ValueError: if user does not provide bounding box#   """#   if bbox is None:#     raise ValueError('Please supply a bounding box.')##   image = decode_jpeg(image_buffer)#   height = FLAGS.image_size#   width = FLAGS.image_size##   if train:#     image = distort_image(image, height, width, bbox, thread_id)#   else:#     image = eval_image(image, height, width)##   # Finally, rescale to [-1,1] instead of [0, 1)#   image = tf.subtract(image, 0.5)#   image = tf.multiply(image, 2.0)#   return image### def parse_example_proto(example_serialized):#   """Parses an Example proto containing a training example of an image.##   The output of the build_image_data.py image preprocessing script is a dataset#   containing serialized Example protocol buffers. Each Example proto contains#   the following fields:##     image/height: 462#     image/width: 581#     image/colorspace: 'RGB'#     image/channels: 3#     image/class/label: 615#     image/class/synset: 'n03623198'#     image/class/text: 'knee pad'#     image/object/bbox/xmin: 0.1#     image/object/bbox/xmax: 0.9#     image/object/bbox/ymin: 0.2#     image/object/bbox/ymax: 0.6#     image/object/bbox/label: 615#     image/format: 'JPEG'#     image/filename: 'ILSVRC2012_val_00041207.JPEG'#     image/encoded: <JPEG encoded string>##   Args:#     example_serialized: scalar Tensor tf.string containing a serialized#       Example protocol buffer.##   Returns:#     image_buffer: Tensor tf.string containing the contents of a JPEG file.#     label: Tensor tf.int32 containing the label.#     bbox: 3-D float Tensor of bounding boxes arranged [1, num_boxes, coords]#       where each coordinate is [0, 1) and the coordinates are arranged as#       [ymin, xmin, ymax, xmax].#     text: Tensor tf.string containing the human-readable label.#   """#   # Dense features in Example proto.#   feature_map = {#       'image/encoded': tf.FixedLenFeature([], dtype=tf.string,#                                           default_value=''),#       'image/class/label': tf.FixedLenFeature([1], dtype=tf.int64,#                                               default_value=-1),#       'image/class/text': tf.FixedLenFeature([], dtype=tf.string,#                                              default_value=''),#   }#   sparse_float32 = tf.VarLenFeature(dtype=tf.float32)#   # Sparse features in Example proto.#   feature_map.update(#       {k: sparse_float32 for k in ['image/object/bbox/xmin',#                                    'image/object/bbox/ymin',#                                    'image/object/bbox/xmax',#                                    'image/object/bbox/ymax']})##   features = tf.parse_single_example(example_serialized, feature_map)#   label = tf.cast(features['image/class/label'], dtype=tf.int32)##   xmin = tf.expand_dims(features['image/object/bbox/xmin'].values, 0)#   ymin = tf.expand_dims(features['image/object/bbox/ymin'].values, 0)#   xmax = tf.expand_dims(features['image/object/bbox/xmax'].values, 0)#   ymax = tf.expand_dims(features['image/object/bbox/ymax'].values, 0)##   # Note that we impose an ordering of (y, x) just to make life difficult.#   bbox = tf.concat(axis=0, values=[ymin, xmin, ymax, xmax])##   # Force the variable number of bounding boxes into the shape#   # [1, num_boxes, coords].#   bbox = tf.expand_dims(bbox, 0)#   bbox = tf.transpose(bbox, [0, 2, 1])##   return features['image/encoded'], label, bbox, features['image/class/text']### def batch_inputs(dataset, batch_size, train, num_preprocess_threads=None,#                  num_readers=1):#   """Contruct batches of training or evaluation examples from the image dataset.##   Args:#     dataset: instance of Dataset class specifying the dataset.#       See dataset.py for details.#     batch_size: integer#     train: boolean#     num_preprocess_threads: integer, total number of preprocessing threads#     num_readers: integer, number of parallel readers##   Returns:#     images: 4-D float Tensor of a batch of images#     labels: 1-D integer Tensor of [batch_size].##   Raises:#     ValueError: if data is not found#   """#   with tf.name_scope('batch_processing'):#     data_files = dataset.data_files()#     if data_files is None:#       raise ValueError('No data files found for this dataset')##     # Create filename_queue#     if train:#       filename_queue = tf.train.string_input_producer(data_files,#                                                       shuffle=True,#                                                       capacity=16)#     else:#       filename_queue = tf.train.string_input_producer(data_files,#                                                       shuffle=False,#                                                       capacity=1)#     if num_preprocess_threads is None:#       num_preprocess_threads = FLAGS.num_preprocess_threads##     if num_preprocess_threads % 4:#       raise ValueError('Please make num_preprocess_threads a multiple '#                        'of 4 (%d % 4 != 0).', num_preprocess_threads)##     if num_readers is None:#       num_readers = FLAGS.num_readers##     if num_readers < 1:#       raise ValueError('Please make num_readers at least 1')##     # Approximate number of examples per shard.#     examples_per_shard = 1024#     # Size the random shuffle queue to balance between good global#     # mixing (more examples) and memory use (fewer examples).#     # 1 image uses 299*299*3*4 bytes = 1MB#     # The default input_queue_memory_factor is 16 implying a shuffling queue#     # size: examples_per_shard * 16 * 1MB = 17.6GB#     min_queue_examples = examples_per_shard * FLAGS.input_queue_memory_factor#     if train:#       examples_queue = tf.RandomShuffleQueue(#           capacity=min_queue_examples + 3 * batch_size,#           min_after_dequeue=min_queue_examples,#           dtypes=[tf.string])#     else:#       examples_queue = tf.FIFOQueue(#           capacity=examples_per_shard + 3 * batch_size,#           dtypes=[tf.string])##     # Create multiple readers to populate the queue of examples.#     if num_readers > 1:#       enqueue_ops = []#       for _ in range(num_readers):#         reader = dataset.reader()#         _, value = reader.read(filename_queue)#         enqueue_ops.append(examples_queue.enqueue([value]))##       tf.train.queue_runner.add_queue_runner(#           tf.train.queue_runner.QueueRunner(examples_queue, enqueue_ops))#       example_serialized = examples_queue.dequeue()#     else:#       reader = dataset.reader()#       _, example_serialized = reader.read(filename_queue)##     images_and_labels = []#     for thread_id in range(num_preprocess_threads):#       # Parse a serialized Example proto to extract the image and metadata.#       image_buffer, label_index, bbox, _ = parse_example_proto(#           example_serialized)#       image = image_preprocessing(image_buffer, bbox, train, thread_id)#       images_and_labels.append([image, label_index])##     images, label_index_batch = tf.train.batch_join(#         images_and_labels,#         batch_size=batch_size,#         capacity=2 * num_preprocess_threads * batch_size)##     # Reshape images into these desired dimensions.#     height = FLAGS.image_size#     width = FLAGS.image_size#     depth = 3##     images = tf.cast(images, tf.float32)#     images = tf.reshape(images, shape=[batch_size, height, width, depth])##     # Display the training images in the visualizer.#     tf.summary.image('images', images)##     return images, tf.reshape(label_index_batch, [batch_size])def inputs(dataset, batch_size=None, num_preprocess_threads=None):  """Generate batches of ImageNet images for evaluation.  Use this function as the inputs for evaluating a network.  Note that some (minimal) image preprocessing occurs during evaluation  including central cropping and resizing of the image to fit the network.  Args:    dataset: instance of Dataset class specifying the dataset.    batch_size: integer, number of examples in batch    num_preprocess_threads: integer, total number of preprocessing threads but      None defaults to FLAGS.num_preprocess_threads.  Returns:    images: Images. 4D tensor of size [batch_size, FLAGS.image_size,                                       image_size, 3].    labels: 1-D integer Tensor of [FLAGS.batch_size].  """  if not batch_size:    batch_size = FLAGS.batch_size  # Force all input processing onto CPU in order to reserve the GPU for  # the forward inference and back-propagation.  with tf.device('/cpu:0'):    images, labels, filenames = batch_inputs(        dataset, batch_size, train=False,        num_preprocess_threads=num_preprocess_threads,        num_readers=1)  return images, labels, filenamesdef distorted_inputs(dataset, batch_size=None, num_preprocess_threads=None):  """Generate batches of distorted versions of ImageNet images.  Use this function as the inputs for training a network.  Distorting images provides a useful technique for augmenting the data  set during training in order to make the network invariant to aspects  of the image that do not effect the label.  Args:    dataset: instance of Dataset class specifying the dataset.    batch_size: integer, number of examples in batch    num_preprocess_threads: integer, total number of preprocessing threads but      None defaults to FLAGS.num_preprocess_threads.  Returns:    images: Images. 4D tensor of size [batch_size, FLAGS.image_size,                                       FLAGS.image_size, 3].    labels: 1-D integer Tensor of [batch_size].  """  if not batch_size:    batch_size = FLAGS.batch_size  # Force all input processing onto CPU in order to reserve the GPU for  # the forward inference and back-propagation.  with tf.device('/cpu:0'):    images, labels, _ = batch_inputs(        dataset, batch_size, train=True,        num_preprocess_threads=num_preprocess_threads,        num_readers=FLAGS.num_readers)  return images, labelsdef decode_jpeg(image_buffer, scope=None):  """Decode a JPEG string into one 3-D float image Tensor.  Args:    image_buffer: scalar string Tensor.    scope: Optional scope for op_scope.  Returns:    3-D float Tensor with values ranging from [0, 1).  """  with tf.op_scope([image_buffer], scope, 'decode_jpeg'):    # Decode the string as an RGB JPEG.    # Note that the resulting image contains an unknown height and width    # that is set dynamically by decode_jpeg. In other words, the height    # and width of image is unknown at compile-time.    image = tf.image.decode_jpeg(image_buffer, channels=3)    # After this point, all image pixels reside in [0,1)    # until the very end, when they're rescaled to (-1, 1).  The various    # adjust_* ops all require this range for dtype float.    image = tf.image.convert_image_dtype(image, dtype=tf.float32)    return imagedef distort_color(image, thread_id=0, scope=None):  """Distort the color of the image.  Each color distortion is non-commutative and thus ordering of the color ops  matters. Ideally we would randomly permute the ordering of the color ops.  Rather then adding that level of complication, we select a distinct ordering  of color ops for each preprocessing thread.  Args:    image: Tensor containing single image.    thread_id: preprocessing thread ID.    scope: Optional scope for op_scope.  Returns:    color-distorted image  """  with tf.op_scope([image], scope, 'distort_color'):    color_ordering = thread_id % 2    if color_ordering == 0:      image = tf.image.random_brightness(image, max_delta=32. / 255.)      image = tf.image.random_saturation(image, lower=0.5, upper=1.5)      image = tf.image.random_hue(image, max_delta=0.2)      image = tf.image.random_contrast(image, lower=0.5, upper=1.5)    elif color_ordering == 1:      image = tf.image.random_brightness(image, max_delta=32. / 255.)      image = tf.image.random_contrast(image, lower=0.5, upper=1.5)      image = tf.image.random_saturation(image, lower=0.5, upper=1.5)      image = tf.image.random_hue(image, max_delta=0.2)    # The random_* ops do not necessarily clamp.    image = tf.clip_by_value(image, 0.0, 1.0)    return imagedef distort_image(image, height, width, bbox, thread_id=0, scope=None):  """Distort one image for training a network.  Distorting images provides a useful technique for augmenting the data  set during training in order to make the network invariant to aspects  of the image that do not effect the label.  Args:    image: 3-D float Tensor of image    height: integer    width: integer    bbox: 3-D float Tensor of bounding boxes arranged [1, num_boxes, coords]      where each coordinate is [0, 1) and the coordinates are arranged      as [ymin, xmin, ymax, xmax].    thread_id: integer indicating the preprocessing thread.    scope: Optional scope for op_scope.  Returns:    3-D float Tensor of distorted image used for training.  """  with tf.op_scope([image, height, width, bbox], scope, 'distort_image'):    # Each bounding box has shape [1, num_boxes, box coords] and    # the coordinates are ordered [ymin, xmin, ymax, xmax].    # Display the bounding box in the first thread only.    if not thread_id:      image_with_box = tf.image.draw_bounding_boxes(tf.expand_dims(image, 0),                                                    bbox)      tf.image_summary('image_with_bounding_boxes', image_with_box)  # A large fraction of image datasets contain a human-annotated bounding  # box delineating the region of the image containing the object of interest.  # We choose to create a new bounding box for the object which is a randomly  # distorted version of the human-annotated bounding box that obeys an allowed  # range of aspect ratios, sizes and overlap with the human-annotated  # bounding box. If no box is supplied, then we assume the bounding box is  # the entire image.    sample_distorted_bounding_box = tf.image.sample_distorted_bounding_box(        tf.shape(image),        bounding_boxes=bbox,        min_object_covered=0.1,        aspect_ratio_range=[0.75, 1.33],        area_range=[0.05, 1.0],        max_attempts=100,        use_image_if_no_bounding_boxes=True)    bbox_begin, bbox_size, distort_bbox = sample_distorted_bounding_box    if not thread_id:      image_with_distorted_box = tf.image.draw_bounding_boxes(          tf.expand_dims(image, 0), distort_bbox)      tf.image_summary('images_with_distorted_bounding_box',                       image_with_distorted_box)    # Crop the image to the specified bounding box.    distorted_image = tf.slice(image, bbox_begin, bbox_size)    # This resizing operation may distort the images because the aspect    # ratio is not respected. We select a resize method in a round robin    # fashion based on the thread number.    # Note that ResizeMethod contains 4 enumerated resizing methods.    resize_method = thread_id % 4    distorted_image = tf.image.resize_images(distorted_image, [height, width],                                             method=resize_method)    # Restore the shape since the dynamic slice based upon the bbox_size loses    # the third dimension.    distorted_image.set_shape([height, width, 3])    if not thread_id:      tf.image_summary('cropped_resized_image',                       tf.expand_dims(distorted_image, 0))    # Randomly flip the image horizontally.    distorted_image = tf.image.random_flip_left_right(distorted_image)    # Randomly distort the colors.    distorted_image = distort_color(distorted_image, thread_id)    if not thread_id:      tf.image_summary('final_distorted_image',                       tf.expand_dims(distorted_image, 0))    return distorted_imagedef eval_image(image, height, width, scope=None):  """Prepare one image for evaluation.  Args:    image: 3-D float Tensor    height: integer    width: integer    scope: Optional scope for op_scope.  Returns:    3-D float Tensor of prepared image.  """  with tf.op_scope([image, height, width], scope, 'eval_image'):    # Crop the central region of the image with an area containing 87.5% of    # the original image.    image = tf.image.central_crop(image, central_fraction=0.875)    # Resize the image to the original height and width.    image = tf.expand_dims(image, 0)    image = tf.image.resize_bilinear(image, [height, width],                                     align_corners=False)    image = tf.squeeze(image, [0])    return imagedef image_preprocessing(image_buffer, bbox, train, thread_id=0):  """Decode and preprocess one image for evaluation or training.  Args:    image_buffer: JPEG encoded string Tensor    bbox: 3-D float Tensor of bounding boxes arranged [1, num_boxes, coords]      where each coordinate is [0, 1) and the coordinates are arranged as      [ymin, xmin, ymax, xmax].    train: boolean    thread_id: integer indicating preprocessing thread  Returns:    3-D float Tensor containing an appropriately scaled image  Raises:    ValueError: if user does not provide bounding box  """  if bbox is None:    raise ValueError('Please supply a bounding box.')  image = decode_jpeg(image_buffer)  height = FLAGS.image_size  width = FLAGS.image_size  if train:    image = distort_image(image, height, width, bbox, thread_id)  else:    image = eval_image(image, height, width)  # Finally, rescale to [-1,1] instead of [0, 1)  image = tf.subtract(image, 0.5)  image = tf.multiply(image, 2.0)  return imagedef debug_print(y):    with tf.Session():     print(y.eval())def parse_example_proto(example_serialized):  """Parses an Example proto containing a training example of an image.  The output of the build_image_data.py image preprocessing script is a dataset  containing serialized Example protocol buffers. Each Example proto contains  the following fields:    image/height: 462    image/width: 581    image/colorspace: 'RGB'    image/channels: 3    image/class/label: 615    image/class/synset: 'n03623198'    image/class/text: 'knee pad'    image/object/bbox/xmin: 0.1    image/object/bbox/xmax: 0.9    image/object/bbox/ymin: 0.2    image/object/bbox/ymax: 0.6    image/object/bbox/label: 615    image/format: 'JPEG'    image/filename: 'ILSVRC2012_val_00041207.JPEG'    image/encoded: <JPEG encoded string>  Args:    example_serialized: scalar Tensor tf.string containing a serialized      Example protocol buffer.  Returns:    image_buffer: Tensor tf.string containing the contents of a JPEG file.    label: Tensor tf.int32 containing the label.    bbox: 3-D float Tensor of bounding boxes arranged [1, num_boxes, coords]      where each coordinate is [0, 1) and the coordinates are arranged as      [ymin, xmin, ymax, xmax].    text: Tensor tf.string containing the human-readable label.  """  # Dense features in Example proto.  feature_map = {      'image/encoded': tf.FixedLenFeature([], dtype=tf.string,                                          default_value=''),      'image/class/label': tf.FixedLenFeature([1], dtype=tf.int64,                                              default_value=-1),      'image/class/text': tf.FixedLenFeature([], dtype=tf.string,                                             default_value=''),      'image/filename': tf.FixedLenFeature([], dtype=tf.string,                                             default_value=''),  }  sparse_float32 = tf.VarLenFeature(dtype=tf.float32)  # Sparse features in Example proto.  feature_map.update(      {k: sparse_float32 for k in ['image/object/bbox/xmin',                                   'image/object/bbox/ymin',                                   'image/object/bbox/xmax',                                   'image/object/bbox/ymax']})  features = tf.parse_single_example(example_serialized, feature_map)  label = tf.cast(features['image/class/label'], dtype=tf.int32)  xmin = tf.expand_dims(features['image/object/bbox/xmin'].values, 0)  ymin = tf.expand_dims(features['image/object/bbox/ymin'].values, 0)  xmax = tf.expand_dims(features['image/object/bbox/xmax'].values, 0)  ymax = tf.expand_dims(features['image/object/bbox/ymax'].values, 0)  # Note that we impose an ordering of (y, x) just to make life difficult.  bbox = tf.concat(axis=0, values=[ymin, xmin, ymax, xmax])  # Force the variable number of bounding boxes into the shape  # [1, num_boxes, coords].  bbox = tf.expand_dims(bbox, 0)  bbox = tf.transpose(bbox, [0, 2, 1])  return features['image/encoded'], label, bbox, features['image/class/text'], features['image/filename']def batch_inputs(dataset, batch_size, train, num_preprocess_threads=None,                 num_readers=1):  """Contruct batches of training or evaluation examples from the image dataset.  Args:    dataset: instance of Dataset class specifying the dataset.      See dataset.py for details.    batch_size: integer    train: boolean    num_preprocess_threads: integer, total number of preprocessing threads    num_readers: integer, number of parallel readers  Returns:    images: 4-D float Tensor of a batch of images    labels: 1-D integer Tensor of [batch_size].    filename list: the list of filename  Raises:    ValueError: if data is not found  """  with tf.name_scope('batch_processing'):    data_files = dataset.data_files()    if data_files is None:      raise ValueError('No data files found for this dataset')    # Create filename_queue    if train:      filename_queue = tf.train.string_input_producer(data_files,                                                      shuffle=True,                                                      capacity=16)    else:      filename_queue = tf.train.string_input_producer(data_files,                                                      shuffle=False,                                                        capacity=1)    if num_preprocess_threads is None:      num_preprocess_threads = FLAGS.num_preprocess_threads    if num_preprocess_threads % 4:      _=1      #raise ValueError('Please make num_preprocess_threads a multiple '      #                 'of 4 (%d % 4 != 0).', num_preprocess_threads)    if num_readers is None:      num_readers = FLAGS.num_readers    if num_readers < 1:      raise ValueError('Please make num_readers at least 1')    # Approximate number of examples per shard.    examples_per_shard = 1024    # Size the random shuffle queue to balance between good global    # mixing (more examples) and memory use (fewer examples).    # 1 image uses 299*299*3*4 bytes = 1MB    # The default input_queue_memory_factor is 16 implying a shuffling queue    # size: examples_per_shard * 16 * 1MB = 17.6GB    min_queue_examples = examples_per_shard * FLAGS.input_queue_memory_factor    if train:      examples_queue = tf.RandomShuffleQueue(          capacity=min_queue_examples + 3 * batch_size,          min_after_dequeue=min_queue_examples,          dtypes=[tf.string])    else:      examples_queue = tf.FIFOQueue(          capacity=examples_per_shard + 3 * batch_size,          dtypes=[tf.string])    # Create multiple readers to populate the queue of examples.    if num_readers > 1:      enqueue_ops = []      for _ in range(num_readers):        reader = dataset.reader()        _, value = reader.read(filename_queue)        enqueue_ops.append(examples_queue.enqueue([value]))      tf.train.queue_runner.add_queue_runner(          tf.train.queue_runner.QueueRunner(examples_queue, enqueue_ops))      example_serialized = examples_queue.dequeue()    else:      reader = dataset.reader()      _, example_serialized = reader.read(filename_queue)    images_and_labels = []    for thread_id in range(num_preprocess_threads):      # Parse a serialized Example proto to extract the image and metadata.      image_buffer, label_index, bbox, _, filename = parse_example_proto(          example_serialized)      image = image_preprocessing(image_buffer, bbox, train, thread_id)      images_and_labels.append([image, label_index,filename])    images, label_index_batch,filenames = tf.train.batch_join(        images_and_labels,        batch_size=batch_size,        capacity=3 * num_preprocess_threads * batch_size)    # Reshape images into these desired dimensions.    height = FLAGS.image_size    width = FLAGS.image_size    depth = 3    images = tf.cast(images, tf.float32)    images = tf.reshape(images, shape=[batch_size, height, width, depth])    # Display the training images in the visualizer.    tf.summary.image('images', images)    return images, tf.reshape(label_index_batch, [batch_size]), tf.reshape(filenames, [batch_size])

这些大概就是这段时间遇到问题的总结  写下来虽然不是很多  但是作为小白 确实花了很多时间去折腾