how to feed Tensorflow dataset pipeline with an example of triple tensors

Question

As it is described in official Tensorflow website, we can feed a dataset pipeline with examples of pair tensor (input, label). I need to know how can I add one more item like ( input, lable1, lable2)?

Answer 1

Simple!

You just make your dataset method output dictionary instead.

This code is from the link you posted, all the way down to the bottom.

def dataset_input_fn():
  filenames = ["/var/data/file1.tfrecord", "/var/data/file2.tfrecord"]
  dataset = tf.data.TFRecordDataset(filenames)

  # Use `tf.parse_single_example()` to extract data from a `tf.Example`
  # protocol buffer, and perform any additional per-record preprocessing.
  def parser(record):
    keys_to_features = {
        "image_data": tf.FixedLenFeature((), tf.string, default_value=""),
        "date_time": tf.FixedLenFeature((), tf.int64, default_value=""),
        "label": tf.FixedLenFeature((), tf.int64,
                                    default_value=tf.zeros([], dtype=tf.int64)),
    }
    parsed = tf.parse_single_example(record, keys_to_features)

    # Perform additional preprocessing on the parsed data.
    image = tf.image.decode_jpeg(parsed["image_data"])
    image = tf.reshape(image, [299, 299, 1])
    label = tf.cast(parsed["label"], tf.int32)

    return {"image_data": image, "date_time": parsed["date_time"]}, label

  # Use `Dataset.map()` to build a pair of a feature dictionary and a label
  # tensor for each example.
  dataset = dataset.map(parser)
  dataset = dataset.shuffle(buffer_size=10000)
  dataset = dataset.batch(32)
  dataset = dataset.repeat(num_epochs)
  iterator = dataset.make_one_shot_iterator()

  # `features` is a dictionary in which each value is a batch of values for
  # that feature; `labels` is a batch of labels.
  features, labels = iterator.get_next()
  return features, labels

Now, features is actually a dictionary with the fields image_data and date_time. This way, you can add as much as you want to either features or labels, while still sticking to the two outputs.