tensorflow – Reading the data – Read & Parse TFRecord file

TFRecord files is the native tensorflow binary format for storing data (tensors). To read the file you can use a code similar to the CSV example:

import tensorflow as tf
filename_queue = tf.train.string_input_producer(["file.tfrecord"], num_epochs=1)
reader = tf.TFRecordReader()
key, serialized_example = reader.read(filename_queue)

Then, you need to parse the examples from serialized_example Queue. You can do it either using tf.parse_example, which requires previous batching, but is faster or tf.parse_single_example:

batch = tf.train.batch([serialized_example], batch_size=100)
parsed_batch = tf.parse_example(batch, features={
  "feature_name_1": tf.FixedLenFeature(shape=[1], tf.int64),
  "feature_name_2": tf.FixedLenFeature(shape=[1], tf.float32)

tf.train.batch joins consecutive values of given tensors of shape [x, y, z] to tensors of shape [batch_size, x, y, z].
features dict maps names of the features to tensorflow’s definitions of features. You use parse_single_example in a similar way:

parsed_example = tf.parse_single_example(serialized_example, {
  "feature_name_1": tf.FixedLenFeature(shape=[1], tf.int64),
  "feature_name_2": tf.FixedLenFeature(shape=[1], tf.float32)

tf.parse_example and tf.parse_single_example return a dictionary mapping feature names to the tensor with the values.

To batch examples coming from parse_single_example you should extract the tensors from the dict and use tf.train.batch as before:

parsed_batch = dict(zip(parsed_example.keys(),
    tf.train.batch(parsed_example.values(), batch_size=100)

You read the data as before, passing the list of all the tensors to evaluate to sess.run:

with tf.Session() as sess:
    while True:
      data_batch = sess.run(parsed_batch.values())
      # process data
  except tf.errors.OutOfRangeError:

if you want to reproduce, please indicate the source:
tensorflow – Reading the data – Read & Parse TFRecord file - CodeDay