How to parse a single TFrecord file

问题

To read tfrecords:

reader = tf.TFRecordReader()
_, serialized_example = reader.read(filename_queue)
features = tf.parse_single_example(...)

TFRecordReader reads examples from a file queue. But how to read a single example from a particular file synchronically(without a queue). like

file_buf = tf.read_file(filename)
serialized_example = get_train_example(file_buf)
features = tf.parse_single_example(...)

how to implement the get_train_example function

回答1:

Not sure if this exactly what you're looking for, but you can do it this way without a queue:

tf_record = "path/to/my.tfrecord"
e = tf.python_io.tf_record_iterator(tf_record).next()
single_example = tf.parse_single_example(e, features=features)

回答2:

used this one

for example in tf.python_io.tf_record_iterator("path_to_your_file"):
    result = tf.train.Example.FromString(example)
print(result)

found it here

来源：https://stackoverflow.com/questions/42946547/how-to-parse-a-single-tfrecord-file