Unable to read a TFRecord file as an Image

TensorFlow General Discussion
,
1

I have a lot of TFRecord images from google earth engine that I am trying to read in as images/maps. I tried to follow this page:

and my code is as follows:

raw_image_dataset = tf.data.TFRecordDataset('uganda_2005_01.tfrecord')

# Create a dictionary describing the features.
image_feature_description = {
    'height': tf.io.FixedLenFeature([], tf.int64),
    'width': tf.io.FixedLenFeature([], tf.int64),
    'depth': tf.io.FixedLenFeature([], tf.int64),
    'label': tf.io.FixedLenFeature([], tf.int64),
    'image_raw': tf.io.FixedLenFeature([], tf.string),
}

def _parse_image_function(example_proto):
  # Parse the input tf.train.Example proto using the dictionary above.
  return tf.io.parse_single_example(example_proto, image_feature_description)

parsed_image_dataset = raw_image_dataset.map(_parse_image_function)
parsed_image_dataset

for image_features in parsed_image_dataset:
  image_raw = image_features['image_raw'].numpy()
  display.display(display.Image(data=image_raw))

I keep getting this error that I am not sure how to fix:

InvalidArgumentError Traceback (most recent call last)
/var/folders/my/mlj39v8d0jl1gjcg98xyly3h0000gn/T/ipykernel_99279/1152503565.py in
----> 1 for image_features in parsed_image_dataset:
2 image_raw = image_features[‘image_raw’].numpy()
3 display.display(display.Image(data=image_raw))

~/opt/miniconda3/envs/CNNRep/lib/python3.7/site-packages/tensorflow_core/python/data/ops/iterator_ops.py in next(self)
620
621 def next(self): # For Python 3 compatibility
→ 622 return self.next()
623
624 def _next_internal(self):

~/opt/miniconda3/envs/CNNRep/lib/python3.7/site-packages/tensorflow_core/python/data/ops/iterator_ops.py in next(self)
664 “”“Returns a nested structure of Tensors containing the next element.”“”
665 try:
→ 666 return self._next_internal()
667 except errors.OutOfRangeError:
668 raise StopIteration

~/opt/miniconda3/envs/CNNRep/lib/python3.7/site-packages/tensorflow_core/python/data/ops/iterator_ops.py in _next_internal(self)
649 self._iterator_resource,
650 output_types=self._flat_output_types,
→ 651 output_shapes=self._flat_output_shapes)
652
653 try:

~/opt/miniconda3/envs/CNNRep/lib/python3.7/site-packages/tensorflow_core/python/ops/gen_dataset_ops.py in iterator_get_next_sync(iterator, output_types, output_shapes, name)
2671 else:
2672 message = e.message
→ 2673 _six.raise_from(_core._status_to_exception(e.code, message), None)
2674 # Add nodes to the TensorFlow graph.
2675 if not isinstance(output_types, (list, tuple)):

~/opt/miniconda3/envs/CNNRep/lib/python3.7/site-packages/six.py in raise_from(value, from_value)

InvalidArgumentError: {{function_node __inference_Dataset_map__parse_image_function_61}} Feature: depth (data type: int64) is required but could not be found.
[[{{node ParseSingleExample/ParseSingleExample}}]] [Op:IteratorGetNextSync]

Any help would be greatly appreciated.

3

Hi @Prerna_Mishra,

This error is due to that the depth feature is not found in the TFRecord file you are trying to parse. To resolve this, you can check the features of the TFRecord once and update your image_feature_description accordingly. I’ve included a sample gist code for checking the features in a sample TFRecord file for your reference.

Thank You.