How do I convert a directory of jpeg images to TFRecords file in tensorflow?

Question

I have training data that is a directory of jpeg images and a corresponding text file containing the file name and the associated category label. I am trying to convert this training data into a tfrecords file as described in the tensorflow documentation. I have spent quite some time trying to get this to work but there are no examples in tensorflow that demonstrate how to use any of the readers to read in jpeg files and add them to a tfrecord using tfrecordwriter

Answer 1

Try this script: (used with VOC segmentation dataset:http://host.robots.ox.ac.uk/pascal/VOC/voc2012/)

import numpy as np
import tensorflow as tf
import scipy.io # to read .mat files
from PIL import Image # to read image files
            
        
    
def get_image(path):
        jpg = Image.open(path).convert('RGB')
        return np.array(jpg)
    
def get_label_png(path):
        png = Image.open(path) # image is saved as palettised png. 
        arr = np.array(png)
        return arr[..., None]
        
def get_example(image, label):
    feature = {
        'height': tf.train.Feature(int64_list=tf.train.Int64List(value=[image.shape[0]])),
        'width': tf.train.Feature(int64_list=tf.train.Int64List(value=[image.shape[1]])),
        'image': tf.train.Feature(bytes_list=tf.train.BytesList(value=[image.tobytes()])),
        'label': tf.train.Feature(bytes_list=tf.train.BytesList(value=[label.tobytes()]))
    }
    return tf.train.Example(features=tf.train.Features(feature=feature))
    
                    
                    
## Paths ======================================
images_folder = 'data/images/' #images folder
labels_folder = 'data/labels/' #label folder
                    
train_file = 'data/train.txt'
val_file = 'data/val.txt'
                    
TRAIN = 'data/train.tfrecords'
VAL = 'data/val.tfrecords'
                    
## write train dataset
with tf.io.TFRecordWriter(TRAIN) as writer:
with open(train_file) as file:
filenames = [s.rstrip('\n') for s in file.readlines()]
for name in filenames:
image = utils.get_image(images_folder+name+'.jpg')
label = utils.get_label_png(labels_folder+name+'.png')
writer.write(utils.get_example(image, label).SerializeToString())
## write validation dataset
with tf.io.TFRecordWriter(VAL) as writer:

 with open(val_file) as file:
  filenames = [s.rstrip('\n') for s in file.readlines()]
  for name in filenames:
      image = utils.get_image(images_folder+name+'.jpg')
      label = utils.get_label_png(labels_folder+name+'.png')
      writer.write(utils.get_example(image, label).SerializeToString())

Answer 2

You can use the Kubeflow pipeline here to do the conversion:

https://aihub.cloud.google.com/u/0/p/products%2Fded3e5e5-d2e8-4d65-9b9f-5ffaa9a27ea1

Click on the Download link (create a Kubeflow cluster to run the pipeline)

Answer 3

In case of too much size in tfrecord files you use directly read bytes.

This link shows it. TFrecords occupy more space than original JPEG images

you use this function to read bytes directly.

img_bytes = open(path,'rb').read()

reference

https://github.com/tensorflow/tensorflow/issues/9675

Answer 4

Mentioning the Code in the Link specified by Kamil, so that the code will be available even if the Link is broken.

"""Converts image data to TFRecords file format with Example protos.

If your data set involves bounding boxes, please look at build_imagenet_data.py.
"""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function

from datetime import datetime
import os
import random
import sys
import threading

import numpy as np
import tensorflow as tf

tf.app.flags.DEFINE_string('train_directory', '/tmp/',
                           'Training data directory')
tf.app.flags.DEFINE_string('validation_directory', '/tmp/',
                           'Validation data directory')
tf.app.flags.DEFINE_string('output_directory', '/tmp/',
                           'Output data directory')

tf.app.flags.DEFINE_integer('train_shards', 2,
                            'Number of shards in training TFRecord files.')
tf.app.flags.DEFINE_integer('validation_shards', 2,
                            'Number of shards in validation TFRecord files.')

tf.app.flags.DEFINE_integer('num_threads', 2,
                            'Number of threads to preprocess the images.')

# The labels file contains a list of valid labels are held in this file.
# Assumes that the file contains entries as such:
#   dog
#   cat
#   flower
# where each line corresponds to a label. We map each label contained in
# the file to an integer corresponding to the line number starting from 0.
tf.app.flags.DEFINE_string('labels_file', '', 'Labels file')


FLAGS = tf.app.flags.FLAGS


def _int64_feature(value):
  """Wrapper for inserting int64 features into Example proto."""
  if not isinstance(value, list):
    value = [value]
  return tf.train.Feature(int64_list=tf.train.Int64List(value=value))


def _bytes_feature(value):
  """Wrapper for inserting bytes features into Example proto."""
  return tf.train.Feature(bytes_list=tf.train.BytesList(value=[value]))


def _convert_to_example(filename, image_buffer, label, text, height, width):
  """Build an Example proto for an example.
  Args:
    filename: string, path to an image file, e.g., '/path/to/example.JPG'
    image_buffer: string, JPEG encoding of RGB image
    label: integer, identifier for the ground truth for the network
    text: string, unique human-readable, e.g. 'dog'
    height: integer, image height in pixels
    width: integer, image width in pixels
  Returns:
    Example proto
  """

  colorspace = 'RGB'
  channels = 3
  image_format = 'JPEG'

  example = tf.train.Example(features=tf.train.Features(feature={
      'image/height': _int64_feature(height),
      'image/width': _int64_feature(width),
      'image/colorspace': _bytes_feature(tf.compat.as_bytes(colorspace)),
      'image/channels': _int64_feature(channels),
      'image/class/label': _int64_feature(label),
      'image/class/text': _bytes_feature(tf.compat.as_bytes(text)),
      'image/format': _bytes_feature(tf.compat.as_bytes(image_format)),
      'image/filename': _bytes_feature(tf.compat.as_bytes(os.path.basename(filename))),
      'image/encoded': _bytes_feature(tf.compat.as_bytes(image_buffer))}))
  return example


class ImageCoder(object):
  """Helper class that provides TensorFlow image coding utilities."""

  def __init__(self):
    # Create a single Session to run all image coding calls.
    self._sess = tf.Session()

    # Initializes function that converts PNG to JPEG data.
    self._png_data = tf.placeholder(dtype=tf.string)
    image = tf.image.decode_png(self._png_data, channels=3)
    self._png_to_jpeg = tf.image.encode_jpeg(image, format='rgb', quality=100)

    # Initializes function that decodes RGB JPEG data.
    self._decode_jpeg_data = tf.placeholder(dtype=tf.string)
    self._decode_jpeg = tf.image.decode_jpeg(self._decode_jpeg_data, channels=3)

  def png_to_jpeg(self, image_data):
    return self._sess.run(self._png_to_jpeg,
                          feed_dict={self._png_data: image_data})

  def decode_jpeg(self, image_data):
    image = self._sess.run(self._decode_jpeg,
                           feed_dict={self._decode_jpeg_data: image_data})
    assert len(image.shape) == 3
    assert image.shape[2] == 3
    return image


def _is_png(filename):
  """Determine if a file contains a PNG format image.
  Args:
    filename: string, path of the image file.
  Returns:
    boolean indicating if the image is a PNG.
  """
  return '.png' in filename


def _process_image(filename, coder):
  """Process a single image file.
  Args:
    filename: string, path to an image file e.g., '/path/to/example.JPG'.
    coder: instance of ImageCoder to provide TensorFlow image coding utils.
  Returns:
    image_buffer: string, JPEG encoding of RGB image.
    height: integer, image height in pixels.
    width: integer, image width in pixels.
  """
  # Read the image file.
  with tf.gfile.FastGFile(filename, 'rb') as f:
    image_data = f.read()

  # Convert any PNG to JPEG's for consistency.
  if _is_png(filename):
    print('Converting PNG to JPEG for %s' % filename)
    image_data = coder.png_to_jpeg(image_data)

  # Decode the RGB JPEG.
  image = coder.decode_jpeg(image_data)

  # Check that image converted to RGB
  assert len(image.shape) == 3
  height = image.shape[0]
  width = image.shape[1]
  assert image.shape[2] == 3

  return image_data, height, width


def _process_image_files_batch(coder, thread_index, ranges, name, filenames,
                               texts, labels, num_shards):
  """Processes and saves list of images as TFRecord in 1 thread.
  Args:
    coder: instance of ImageCoder to provide TensorFlow image coding utils.
    thread_index: integer, unique batch to run index is within [0, len(ranges)).
    ranges: list of pairs of integers specifying ranges of each batches to
      analyze in parallel.
    name: string, unique identifier specifying the data set
    filenames: list of strings; each string is a path to an image file
    texts: list of strings; each string is human readable, e.g. 'dog'
    labels: list of integer; each integer identifies the ground truth
    num_shards: integer number of shards for this data set.
  """
  # Each thread produces N shards where N = int(num_shards / num_threads).
  # For instance, if num_shards = 128, and the num_threads = 2, then the first
  # thread would produce shards [0, 64).
  num_threads = len(ranges)
  assert not num_shards % num_threads
  num_shards_per_batch = int(num_shards / num_threads)

  shard_ranges = np.linspace(ranges[thread_index][0],
                             ranges[thread_index][1],
                             num_shards_per_batch + 1).astype(int)
  num_files_in_thread = ranges[thread_index][1] - ranges[thread_index][0]

  counter = 0
  for s in range(num_shards_per_batch):
    # Generate a sharded version of the file name, e.g. 'train-00002-of-00010'
    shard = thread_index * num_shards_per_batch + s
    output_filename = '%s-%.5d-of-%.5d' % (name, shard, num_shards)
    output_file = os.path.join(FLAGS.output_directory, output_filename)
    writer = tf.python_io.TFRecordWriter(output_file)

    shard_counter = 0
    files_in_shard = np.arange(shard_ranges[s], shard_ranges[s + 1], dtype=int)
    for i in files_in_shard:
      filename = filenames[i]
      label = labels[i]
      text = texts[i]

      try:
        image_buffer, height, width = _process_image(filename, coder)
      except Exception as e:
        print(e)
        print('SKIPPED: Unexpected eror while decoding %s.' % filename)
        continue

      example = _convert_to_example(filename, image_buffer, label,
                                    text, height, width)
      writer.write(example.SerializeToString())
      shard_counter += 1
      counter += 1

      if not counter % 1000:
        print('%s [thread %d]: Processed %d of %d images in thread batch.' %
              (datetime.now(), thread_index, counter, num_files_in_thread))
        sys.stdout.flush()

    writer.close()
    print('%s [thread %d]: Wrote %d images to %s' %
          (datetime.now(), thread_index, shard_counter, output_file))
    sys.stdout.flush()
    shard_counter = 0
  print('%s [thread %d]: Wrote %d images to %d shards.' %
        (datetime.now(), thread_index, counter, num_files_in_thread))
  sys.stdout.flush()


def _process_image_files(name, filenames, texts, labels, num_shards):
  """Process and save list of images as TFRecord of Example protos.
  Args:
    name: string, unique identifier specifying the data set
    filenames: list of strings; each string is a path to an image file
    texts: list of strings; each string is human readable, e.g. 'dog'
    labels: list of integer; each integer identifies the ground truth
    num_shards: integer number of shards for this data set.
  """
  assert len(filenames) == len(texts)
  assert len(filenames) == len(labels)

  # Break all images into batches with a [ranges[i][0], ranges[i][1]].
  spacing = np.linspace(0, len(filenames), FLAGS.num_threads + 1).astype(np.int)
  ranges = []
  for i in range(len(spacing) - 1):
    ranges.append([spacing[i], spacing[i + 1]])

  # Launch a thread for each batch.
  print('Launching %d threads for spacings: %s' % (FLAGS.num_threads, ranges))
  sys.stdout.flush()

  # Create a mechanism for monitoring when all threads are finished.
  coord = tf.train.Coordinator()

  # Create a generic TensorFlow-based utility for converting all image codings.
  coder = ImageCoder()

  threads = []
  for thread_index in range(len(ranges)):
    args = (coder, thread_index, ranges, name, filenames,
            texts, labels, num_shards)
    t = threading.Thread(target=_process_image_files_batch, args=args)
    t.start()
    threads.append(t)

  # Wait for all the threads to terminate.
  coord.join(threads)
  print('%s: Finished writing all %d images in data set.' %
        (datetime.now(), len(filenames)))
  sys.stdout.flush()


def _find_image_files(data_dir, labels_file):
  """Build a list of all images files and labels in the data set.
  Args:
    data_dir: string, path to the root directory of images.
      Assumes that the image data set resides in JPEG files located in
      the following directory structure.
        data_dir/dog/another-image.JPEG
        data_dir/dog/my-image.jpg
      where 'dog' is the label associated with these images.
    labels_file: string, path to the labels file.
      The list of valid labels are held in this file. Assumes that the file
      contains entries as such:
        dog
        cat
        flower
      where each line corresponds to a label. We map each label contained in
      the file to an integer starting with the integer 0 corresponding to the
      label contained in the first line.
  Returns:
    filenames: list of strings; each string is a path to an image file.
    texts: list of strings; each string is the class, e.g. 'dog'
    labels: list of integer; each integer identifies the ground truth.
  """
  print('Determining list of input files and labels from %s.' % data_dir)
  unique_labels = [l.strip() for l in tf.gfile.FastGFile(
      labels_file, 'r').readlines()]

  labels = []
  filenames = []
  texts = []

  # Leave label index 0 empty as a background class.
  label_index = 1

  # Construct the list of JPEG files and labels.
  for text in unique_labels:
    jpeg_file_path = '%s/%s/*' % (data_dir, text)
    matching_files = tf.gfile.Glob(jpeg_file_path)

    labels.extend([label_index] * len(matching_files))
    texts.extend([text] * len(matching_files))
    filenames.extend(matching_files)

    if not label_index % 100:
      print('Finished finding files in %d of %d classes.' % (
          label_index, len(labels)))
    label_index += 1

  # Shuffle the ordering of all image files in order to guarantee
  # random ordering of the images with respect to label in the
  # saved TFRecord files. Make the randomization repeatable.
  shuffled_index = list(range(len(filenames)))
  random.seed(12345)
  random.shuffle(shuffled_index)

  filenames = [filenames[i] for i in shuffled_index]
  texts = [texts[i] for i in shuffled_index]
  labels = [labels[i] for i in shuffled_index]

  print('Found %d JPEG files across %d labels inside %s.' %
        (len(filenames), len(unique_labels), data_dir))
  return filenames, texts, labels


def _process_dataset(name, directory, num_shards, labels_file):
  """Process a complete data set and save it as a TFRecord.
  Args:
    name: string, unique identifier specifying the data set.
    directory: string, root path to the data set.
    num_shards: integer number of shards for this data set.
    labels_file: string, path to the labels file.
  """
  filenames, texts, labels = _find_image_files(directory, labels_file)
  _process_image_files(name, filenames, texts, labels, num_shards)


def main(unused_argv):
  assert not FLAGS.train_shards % FLAGS.num_threads, (
      'Please make the FLAGS.num_threads commensurate with FLAGS.train_shards')
  assert not FLAGS.validation_shards % FLAGS.num_threads, (
      'Please make the FLAGS.num_threads commensurate with '
      'FLAGS.validation_shards')
  print('Saving results to %s' % FLAGS.output_directory)

  # Run it!
  _process_dataset('validation', FLAGS.validation_directory,
                   FLAGS.validation_shards, FLAGS.labels_file)
  _process_dataset('train', FLAGS.train_directory,
                   FLAGS.train_shards, FLAGS.labels_file)


if __name__ == '__main__':
  tf.app.run()

Answer 5

I have same problem, too.

So here is how i get the tfrecords files of my own jpeg files

Edit: add sol 1 - a better & faster way update: Jan/5/2020

(Recommended) Solution 1: TFRecordWriter

See this Tfrecords Guide post

Solution 2:

From tensorflow official github: How to Construct a New Dataset for Retraining, use official python script build_image_data.py directly and bazel is a better idea.

Here is the instruction:

To run build_image_data.py, you can run the following command line:

# location to where to save the TFRecord data.        
OUTPUT_DIRECTORY=$HOME/my-custom-data/

# build the preprocessing script.
bazel build inception/build_image_data

# convert the data.
bazel-bin/inception/build_image_data \
  --train_directory="${TRAIN_DIR}" \
  --validation_directory="${VALIDATION_DIR}" \
  --output_directory="${OUTPUT_DIRECTORY}" \
  --labels_file="${LABELS_FILE}" \
  --train_shards=128 \
  --validation_shards=24 \
  --num_threads=8

where the $OUTPUT_DIRECTORY is the location of the sharded TFRecords. The $LABELS_FILE will be a text file that is read by the script that provides a list of all of the labels.

then, it should do the trick.

ps. bazel, which is made by Google, turn code into makefile.

Solution 3:

First, i reference the instruction by @capitalistpug and check the shell script file

(shell script file providing by Google: download_and_preprocess_flowers.sh)

Second, i also find out a mini inception-v3 training tutorial by NVIDIA

(NVIDIA official SPEED UP TRAINING WITH GPU-ACCELERATED TENSORFLOW)

Be careful, the following steps need to be executed in the Bazel WORKSAPCE enviroment

so Bazel build file can run successfully

---

First step, I comment out the part of downloading the imagenet data set that i already downloaded

and the rest of the part that i don't need of download_and_preprocess_flowers.sh

Second step, change directory to tensorflow/models/inception

where it is the Bazel environment and it is build by Bazel before

$ cd tensorflow/models/inception 

Optional : If it is not builded before, type in the following code in cmd

$ bazel build inception/download_and_preprocess_flowers 

You need to figure out the content in the following image

And last step, type in the following code:

$ bazel-bin/inception/download_and_preprocess_flowers $Your/own/image/data/path

Then, it will start calling build_image_data.py and creating tfrecords file

Answer 6

I hope this helps:

filename_queue = tf.train.string_input_producer(['/Users/HANEL/Desktop/tf.png']) #  list of files to read

reader = tf.WholeFileReader()
key, value = reader.read(filename_queue)

my_img = tf.image.decode_png(value) # use decode_png or decode_jpeg decoder based on your files.

init_op = tf.initialize_all_variables()
with tf.Session() as sess:
  sess.run(init_op)

# Start populating the filename queue.

coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(coord=coord)

for i in range(1): #length of your filename list
  image = my_img.eval() #here is your image Tensor :) 

print(image.shape)
Image.show(Image.fromarray(np.asarray(image)))

coord.request_stop()
coord.join(threads)

For getting all images as an array of tensors use the following code example.

Github repo of ImageFlow

---

Update:

In the previous answer I just told how to read an image in TF format, but not saving it in TFRecords. For that you should use:

def _int64_feature(value):
  return tf.train.Feature(int64_list=tf.train.Int64List(value=[value]))


def _bytes_feature(value):
  return tf.train.Feature(bytes_list=tf.train.BytesList(value=[value]))

# images and labels array as input
def convert_to(images, labels, name):
  num_examples = labels.shape[0]
  if images.shape[0] != num_examples:
    raise ValueError("Images size %d does not match label size %d." %
                     (images.shape[0], num_examples))
  rows = images.shape[1]
  cols = images.shape[2]
  depth = images.shape[3]

  filename = os.path.join(FLAGS.directory, name + '.tfrecords')
  print('Writing', filename)
  writer = tf.python_io.TFRecordWriter(filename)
  for index in range(num_examples):
    image_raw = images[index].tostring()
    example = tf.train.Example(features=tf.train.Features(feature={
        'height': _int64_feature(rows),
        'width': _int64_feature(cols),
        'depth': _int64_feature(depth),
        'label': _int64_feature(int(labels[index])),
        'image_raw': _bytes_feature(image_raw)}))
    writer.write(example.SerializeToString())

More info here

And you read the data like this:

# Remember to generate a file name queue of you 'train.TFRecord' file path
def read_and_decode(filename_queue):
  reader = tf.TFRecordReader()
  _, serialized_example = reader.read(filename_queue)
  features = tf.parse_single_example(
    serialized_example,
    dense_keys=['image_raw', 'label'],
    # Defaults are not specified since both keys are required.
    dense_types=[tf.string, tf.int64])

  # Convert from a scalar string tensor (whose single string has
  image = tf.decode_raw(features['image_raw'], tf.uint8)

  image = tf.reshape(image, [my_cifar.n_input])
  image.set_shape([my_cifar.n_input])

  # OPTIONAL: Could reshape into a 28x28 image and apply distortions
  # here.  Since we are not applying any distortions in this
  # example, and the next step expects the image to be flattened
  # into a vector, we don't bother.

  # Convert from [0, 255] -> [-0.5, 0.5] floats.
  image = tf.cast(image, tf.float32)
  image = tf.cast(image, tf.float32) * (1. / 255) - 0.5

  # Convert label from a scalar uint8 tensor to an int32 scalar.
  label = tf.cast(features['label'], tf.int32)

  return image, label

Answer 7

Note that images will be saved in TFRecord as uncompressed tensors, possibly increasing the size by a factor of about 5. That's wasting storage space, and likely to be rather slow because of the amount of data that needs to be read.

It's far better to just save the filename in the TFRecord, and read the file on demand. The new Dataset API works well, and the documentation has this example:

# Reads an image from a file, decodes it into a dense tensor, and resizes it
# to a fixed shape.
def _parse_function(filename, label):
  image_string = tf.read_file(filename)
  image_decoded = tf.image.decode_jpeg(image_string)
  image_resized = tf.image.resize_images(image_decoded, [28, 28])
  return image_resized, label

# A vector of filenames.
filenames = tf.constant(["/var/data/image1.jpg", "/var/data/image2.jpg", ...])

# `labels[i]` is the label for the image in `filenames[i].
labels = tf.constant([0, 37, ...])

dataset = tf.data.Dataset.from_tensor_slices((filenames, labels))
dataset = dataset.map(_parse_function)

Answer 8

Tensorflow's inception model has a file build_image_data.py that can accomplish the same thing with the assumption that each subdirectory represents a label.