Merge remote-tracking branch 'upstream/master' into newavarecords

research/attention_ocr/python/metrics.py

@@ -34,20 +34,21 @@ def char_accuracy(predictions, targets, rej_char, streaming=False):
     a update_ops for execution and value tensor whose value on evaluation
     returns the total character accuracy.
   """
-  with tf.variable_scope('CharAccuracy'):
+  with tf.compat.v1.variable_scope('CharAccuracy'):
     predictions.get_shape().assert_is_compatible_with(targets.get_shape())
 
-    targets = tf.to_int32(targets)
+    targets = tf.cast(targets, dtype=tf.int32)
     const_rej_char = tf.constant(rej_char, shape=targets.get_shape())
-    weights = tf.to_float(tf.not_equal(targets, const_rej_char))
-    correct_chars = tf.to_float(tf.equal(predictions, targets))
-    accuracy_per_example = tf.div(
-        tf.reduce_sum(tf.multiply(correct_chars, weights), 1),
-        tf.reduce_sum(weights, 1))
+    weights = tf.cast(tf.not_equal(targets, const_rej_char), dtype=tf.float32)
+    correct_chars = tf.cast(tf.equal(predictions, targets), dtype=tf.float32)
+    accuracy_per_example = tf.compat.v1.div(
+        tf.reduce_sum(input_tensor=tf.multiply(
+            correct_chars, weights), axis=1),
+        tf.reduce_sum(input_tensor=weights, axis=1))
     if streaming:
       return tf.contrib.metrics.streaming_mean(accuracy_per_example)
     else:
-      return tf.reduce_mean(accuracy_per_example)
+      return tf.reduce_mean(input_tensor=accuracy_per_example)
 
 
 def sequence_accuracy(predictions, targets, rej_char, streaming=False):
@@ -66,25 +67,26 @@ def sequence_accuracy(predictions, targets, rej_char, streaming=False):
     returns the total sequence accuracy.
   """
 
-  with tf.variable_scope('SequenceAccuracy'):
+  with tf.compat.v1.variable_scope('SequenceAccuracy'):
     predictions.get_shape().assert_is_compatible_with(targets.get_shape())
 
-    targets = tf.to_int32(targets)
+    targets = tf.cast(targets, dtype=tf.int32)
     const_rej_char = tf.constant(
         rej_char, shape=targets.get_shape(), dtype=tf.int32)
     include_mask = tf.not_equal(targets, const_rej_char)
-    include_predictions = tf.to_int32(
-        tf.where(include_mask, predictions,
-                 tf.zeros_like(predictions) + rej_char))
-    correct_chars = tf.to_float(tf.equal(include_predictions, targets))
+    include_predictions = tf.cast(
+        tf.compat.v1.where(include_mask, predictions,
+                           tf.zeros_like(predictions) + rej_char), dtype=tf.int32)
+    correct_chars = tf.cast(
+        tf.equal(include_predictions, targets), dtype=tf.float32)
     correct_chars_counts = tf.cast(
-        tf.reduce_sum(correct_chars, reduction_indices=[1]), dtype=tf.int32)
+        tf.reduce_sum(input_tensor=correct_chars, axis=[1]), dtype=tf.int32)
     target_length = targets.get_shape().dims[1].value
     target_chars_counts = tf.constant(
         target_length, shape=correct_chars_counts.get_shape())
-    accuracy_per_example = tf.to_float(
-        tf.equal(correct_chars_counts, target_chars_counts))
+    accuracy_per_example = tf.cast(
+        tf.equal(correct_chars_counts, target_chars_counts), dtype=tf.float32)
     if streaming:
       return tf.contrib.metrics.streaming_mean(accuracy_per_example)
     else:
-      return tf.reduce_mean(accuracy_per_example)
+      return tf.reduce_mean(input_tensor=accuracy_per_example)

research/attention_ocr/python/metrics_test.py

@@ -38,8 +38,8 @@ class AccuracyTest(tf.test.TestCase):
       A session object that should be used as a context manager.
     """
     with self.cached_session() as sess:
-      sess.run(tf.global_variables_initializer())
-      sess.run(tf.local_variables_initializer())
+      sess.run(tf.compat.v1.global_variables_initializer())
+      sess.run(tf.compat.v1.local_variables_initializer())
       yield sess
 
   def _fake_labels(self):
@@ -55,7 +55,7 @@ class AccuracyTest(tf.test.TestCase):
     return incorrect
 
   def test_sequence_accuracy_identical_samples(self):
-    labels_tf = tf.convert_to_tensor(self._fake_labels())
+    labels_tf = tf.convert_to_tensor(value=self._fake_labels())
 
     accuracy_tf = metrics.sequence_accuracy(labels_tf, labels_tf,
                                             self.rej_char)
@@ -66,9 +66,9 @@ class AccuracyTest(tf.test.TestCase):
 
   def test_sequence_accuracy_one_char_difference(self):
     ground_truth_np = self._fake_labels()
-    ground_truth_tf = tf.convert_to_tensor(ground_truth_np)
+    ground_truth_tf = tf.convert_to_tensor(value=ground_truth_np)
     prediction_tf = tf.convert_to_tensor(
-        self._incorrect_copy(ground_truth_np, bad_indexes=((0, 0))))
+        value=self._incorrect_copy(ground_truth_np, bad_indexes=((0, 0))))
 
     accuracy_tf = metrics.sequence_accuracy(prediction_tf, ground_truth_tf,
                                             self.rej_char)
@@ -80,9 +80,9 @@ class AccuracyTest(tf.test.TestCase):
 
   def test_char_accuracy_one_char_difference_with_padding(self):
     ground_truth_np = self._fake_labels()
-    ground_truth_tf = tf.convert_to_tensor(ground_truth_np)
+    ground_truth_tf = tf.convert_to_tensor(value=ground_truth_np)
     prediction_tf = tf.convert_to_tensor(
-        self._incorrect_copy(ground_truth_np, bad_indexes=((0, 0))))
+        value=self._incorrect_copy(ground_truth_np, bad_indexes=((0, 0))))
 
     accuracy_tf = metrics.char_accuracy(prediction_tf, ground_truth_tf,
                                         self.rej_char)

research/attention_ocr/python/model.py

@@ -12,7 +12,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 # ==============================================================================
-
 """Functions to build the Attention OCR model.
 
 Usage example:
@@ -26,6 +25,7 @@ Usage example:
 import sys
 import collections
 import logging
+import numpy as np
 import tensorflow as tf
 from tensorflow.contrib import slim
 from tensorflow.contrib.slim.nets import inception
@@ -35,29 +35,28 @@ import sequence_layers
 import utils
 
 OutputEndpoints = collections.namedtuple('OutputEndpoints', [
-  'chars_logit', 'chars_log_prob', 'predicted_chars', 'predicted_scores',
-  'predicted_text'
+    'chars_logit', 'chars_log_prob', 'predicted_chars', 'predicted_scores',
+    'predicted_text', 'predicted_length', 'predicted_conf',
+    'normalized_seq_conf'
 ])
 
 # TODO(gorban): replace with tf.HParams when it is released.
-ModelParams = collections.namedtuple('ModelParams', [
-  'num_char_classes', 'seq_length', 'num_views', 'null_code'
-])
+ModelParams = collections.namedtuple(
+    'ModelParams', ['num_char_classes', 'seq_length', 'num_views', 'null_code'])
 
 ConvTowerParams = collections.namedtuple('ConvTowerParams', ['final_endpoint'])
 
 SequenceLogitsParams = collections.namedtuple('SequenceLogitsParams', [
-  'use_attention', 'use_autoregression', 'num_lstm_units', 'weight_decay',
-  'lstm_state_clip_value'
+    'use_attention', 'use_autoregression', 'num_lstm_units', 'weight_decay',
+    'lstm_state_clip_value'
 ])
 
-SequenceLossParams = collections.namedtuple('SequenceLossParams', [
-  'label_smoothing', 'ignore_nulls', 'average_across_timesteps'
-])
+SequenceLossParams = collections.namedtuple(
+    'SequenceLossParams',
+    ['label_smoothing', 'ignore_nulls', 'average_across_timesteps'])
 
-EncodeCoordinatesParams = collections.namedtuple('EncodeCoordinatesParams', [
-  'enabled'
-])
+EncodeCoordinatesParams = collections.namedtuple('EncodeCoordinatesParams',
+                                                 ['enabled'])
 
 
 def _dict_to_array(id_to_char, default_character):
@@ -85,16 +84,16 @@ class CharsetMapper(object):
     """
     mapping_strings = tf.constant(_dict_to_array(charset, default_character))
     self.table = tf.contrib.lookup.index_to_string_table_from_tensor(
-      mapping=mapping_strings, default_value=default_character)
+        mapping=mapping_strings, default_value=default_character)
 
   def get_text(self, ids):
     """Returns a string corresponding to a sequence of character ids.
 
         Args:
           ids: a tensor with shape [batch_size, max_sequence_length]
-        """
-    return tf.reduce_join(
-      self.table.lookup(tf.to_int64(ids)), reduction_indices=1)
+    """
+    return tf.strings.reduce_join(
+        inputs=self.table.lookup(tf.cast(ids, dtype=tf.int64)), axis=1)
 
 
 def get_softmax_loss_fn(label_smoothing):
@@ -111,16 +110,153 @@ def get_softmax_loss_fn(label_smoothing):
 
     def loss_fn(labels, logits):
       return (tf.nn.softmax_cross_entropy_with_logits(
-        logits=logits, labels=labels))
+          logits=logits, labels=tf.stop_gradient(labels)))
   else:
 
     def loss_fn(labels, logits):
       return tf.nn.sparse_softmax_cross_entropy_with_logits(
-        logits=logits, labels=labels)
+          logits=logits, labels=labels)
 
   return loss_fn
 
 
+def get_tensor_dimensions(tensor):
+  """Returns the shape components of a 4D tensor with variable batch size.
+
+  Args:
+    tensor : A 4D tensor, whose last 3 dimensions are known at graph
+      construction time.
+
+  Returns:
+    batch_size : The first dimension as a tensor object.
+    height : The second dimension as a scalar value.
+    width : The third dimension as a scalar value.
+    num_features : The forth dimension as a scalar value.
+
+  Raises:
+    ValueError: if input tensor does not have 4 dimensions.
+  """
+  if len(tensor.get_shape().dims) != 4:
+    raise ValueError(
+        'Incompatible shape: len(tensor.get_shape().dims) != 4 (%d != 4)' %
+        len(tensor.get_shape().dims))
+  batch_size = tf.shape(input=tensor)[0]
+  height = tensor.get_shape().dims[1].value
+  width = tensor.get_shape().dims[2].value
+  num_features = tensor.get_shape().dims[3].value
+  return batch_size, height, width, num_features
+
+
+def lookup_indexed_value(indices, row_vecs):
+  """Lookup values in each row of 'row_vecs' indexed by 'indices'.
+
+  For each sample in the batch, look up the element for the corresponding
+  index.
+
+  Args:
+    indices : A tensor of shape (batch, )
+    row_vecs : A tensor of shape [batch, depth]
+
+  Returns:
+    A tensor of shape (batch, ) formed by row_vecs[i, indices[i]].
+  """
+  gather_indices = tf.stack((tf.range(
+      tf.shape(input=row_vecs)[0], dtype=tf.int32), tf.cast(indices, tf.int32)),
+      axis=1)
+  return tf.gather_nd(row_vecs, gather_indices)
+
+
+@utils.ConvertAllInputsToTensors
+def max_char_logprob_cumsum(char_log_prob):
+  """Computes the cumulative sum of character logprob for all sequence lengths.
+
+  Args:
+    char_log_prob: A tensor of shape [batch x seq_length x num_char_classes]
+      with log probabilities of a character.
+
+  Returns:
+    A tensor of shape [batch x (seq_length+1)] where each element x[_, j] is
+    the sum of the max char logprob for all positions upto j.
+    Note this duplicates the final column and produces (seq_length+1) columns
+    so the same function can be used regardless whether use_length_predictions
+    is true or false.
+  """
+  max_char_log_prob = tf.reduce_max(input_tensor=char_log_prob, axis=2)
+  # For an input array [a, b, c]) tf.cumsum returns [a, a + b, a + b + c] if
+  # exclusive set to False (default).
+  return tf.cumsum(max_char_log_prob, axis=1, exclusive=False)
+
+
+def find_length_by_null(predicted_chars, null_code):
+  """Determine sequence length by finding null_code among predicted char IDs.
+
+  Given the char class ID for each position, compute the sequence length.
+  Note that this function computes this based on the number of null_code,
+  instead of the position of the first null_code.
+
+  Args:
+    predicted_chars: A tensor of [batch x seq_length] where each element stores
+      the char class ID with max probability;
+    null_code: an int32, character id for the NULL.
+
+  Returns:
+    A [batch, ] tensor which stores the sequence length for each sample.
+  """
+  return tf.reduce_sum(
+      input_tensor=tf.cast(tf.not_equal(null_code, predicted_chars), tf.int32), axis=1)
+
+
+def axis_pad(tensor, axis, before=0, after=0, constant_values=0.0):
+  """Pad a tensor with the specified values along a single axis.
+
+  Args:
+    tensor: a Tensor;
+    axis: the dimension to add pad along to;
+    before: number of values to add before the contents of tensor in the
+      selected dimension;
+    after: number of values to add after the contents of tensor in the selected
+      dimension;
+    constant_values: the scalar pad value to use. Must be same type as tensor.
+
+  Returns:
+    A Tensor. Has the same type as the input tensor, but with a changed shape
+    along the specified dimension.
+  """
+  if before == 0 and after == 0:
+    return tensor
+  ndims = tensor.shape.ndims
+  padding_size = np.zeros((ndims, 2), dtype='int32')
+  padding_size[axis] = before, after
+  return tf.pad(
+      tensor=tensor,
+      paddings=tf.constant(padding_size),
+      constant_values=constant_values)
+
+
+def null_based_length_prediction(chars_log_prob, null_code):
+  """Computes length and confidence of prediction based on positions of NULLs.
+
+  Args:
+    chars_log_prob: A tensor of shape [batch x seq_length x num_char_classes]
+      with log probabilities of a character;
+    null_code: an int32, character id for the NULL.
+
+  Returns:
+    A tuple (text_log_prob, predicted_length), where
+    text_log_prob - is a tensor of the same shape as length_log_prob.
+    Element #0 of the output corresponds to probability of the empty string,
+    element #seq_length - is the probability of length=seq_length.
+    predicted_length is a tensor with shape [batch].
+  """
+  predicted_chars = tf.cast(
+      tf.argmax(input=chars_log_prob, axis=2), dtype=tf.int32)
+  # We do right pad to support sequences with seq_length elements.
+  text_log_prob = max_char_logprob_cumsum(
+      axis_pad(chars_log_prob, axis=1, after=1))
+  predicted_length = find_length_by_null(predicted_chars, null_code)
+  return text_log_prob, predicted_length
+
+
 class Model(object):
   """Class to create the Attention OCR Model."""
 
@@ -137,24 +273,24 @@ class Model(object):
       num_char_classes: size of character set.
       seq_length: number of characters in a sequence.
       num_views: Number of views (conv towers) to use.
-      null_code: A character code corresponding to a character which
-        indicates end of a sequence.
-      mparams: a dictionary with hyper parameters for methods,  keys -
-        function names, values - corresponding namedtuples.
+      null_code: A character code corresponding to a character which indicates
+        end of a sequence.
+      mparams: a dictionary with hyper parameters for methods,  keys - function
+        names, values - corresponding namedtuples.
       charset: an optional dictionary with a mapping between character ids and
-        utf8 strings. If specified the OutputEndpoints.predicted_text will
-        utf8 encoded strings corresponding to the character ids returned by
+        utf8 strings. If specified the OutputEndpoints.predicted_text will utf8
+        encoded strings corresponding to the character ids returned by
         OutputEndpoints.predicted_chars (by default the predicted_text contains
-        an empty vector). 
+        an empty vector).
         NOTE: Make sure you call tf.tables_initializer().run() if the charset
-        specified.
+          specified.
     """
     super(Model, self).__init__()
     self._params = ModelParams(
-      num_char_classes=num_char_classes,
-      seq_length=seq_length,
-      num_views=num_views,
-      null_code=null_code)
+        num_char_classes=num_char_classes,
+        seq_length=seq_length,
+        num_views=num_views,
+        null_code=null_code)
     self._mparams = self.default_mparams()
     if mparams:
       self._mparams.update(mparams)
@@ -162,21 +298,22 @@ class Model(object):
 
   def default_mparams(self):
     return {
-      'conv_tower_fn':
-        ConvTowerParams(final_endpoint='Mixed_5d'),
-      'sequence_logit_fn':
-        SequenceLogitsParams(
-          use_attention=True,
-          use_autoregression=True,
-          num_lstm_units=256,
-          weight_decay=0.00004,
-          lstm_state_clip_value=10.0),
-      'sequence_loss_fn':
-        SequenceLossParams(
-          label_smoothing=0.1,
-          ignore_nulls=True,
-          average_across_timesteps=False),
-      'encode_coordinates_fn': EncodeCoordinatesParams(enabled=False)
+        'conv_tower_fn':
+            ConvTowerParams(final_endpoint='Mixed_5d'),
+        'sequence_logit_fn':
+            SequenceLogitsParams(
+                use_attention=True,
+                use_autoregression=True,
+                num_lstm_units=256,
+                weight_decay=0.00004,
+                lstm_state_clip_value=10.0),
+        'sequence_loss_fn':
+            SequenceLossParams(
+                label_smoothing=0.1,
+                ignore_nulls=True,
+                average_across_timesteps=False),
+        'encode_coordinates_fn':
+            EncodeCoordinatesParams(enabled=False)
     }
 
   def set_mparam(self, function, **kwargs):
@@ -198,14 +335,14 @@ class Model(object):
     """
     mparams = self._mparams['conv_tower_fn']
     logging.debug('Using final_endpoint=%s', mparams.final_endpoint)
-    with tf.variable_scope('conv_tower_fn/INCE'):
+    with tf.compat.v1.variable_scope('conv_tower_fn/INCE'):
       if reuse:
-        tf.get_variable_scope().reuse_variables()
+        tf.compat.v1.get_variable_scope().reuse_variables()
       with slim.arg_scope(inception.inception_v3_arg_scope()):
         with slim.arg_scope([slim.batch_norm, slim.dropout],
                             is_training=is_training):
           net, _ = inception.inception_v3_base(
-            images, final_endpoint=mparams.final_endpoint)
+              images, final_endpoint=mparams.final_endpoint)
       return net
 
   def _create_lstm_inputs(self, net):
@@ -222,10 +359,10 @@ class Model(object):
     """
     num_features = net.get_shape().dims[1].value
     if num_features < self._params.seq_length:
-      raise AssertionError('Incorrect dimension #1 of input tensor'
-                           ' %d should be bigger than %d (shape=%s)' %
-                           (num_features, self._params.seq_length,
-                            net.get_shape()))
+      raise AssertionError(
+          'Incorrect dimension #1 of input tensor'
+          ' %d should be bigger than %d (shape=%s)' %
+          (num_features, self._params.seq_length, net.get_shape()))
     elif num_features > self._params.seq_length:
       logging.warning('Ignoring some features: use %d of %d (shape=%s)',
                       self._params.seq_length, num_features, net.get_shape())
@@ -236,7 +373,7 @@ class Model(object):
   def sequence_logit_fn(self, net, labels_one_hot):
     mparams = self._mparams['sequence_logit_fn']
     # TODO(gorban): remove /alias suffixes from the scopes.
-    with tf.variable_scope('sequence_logit_fn/SQLR'):
+    with tf.compat.v1.variable_scope('sequence_logit_fn/SQLR'):
       layer_class = sequence_layers.get_layer_class(mparams.use_attention,
                                                     mparams.use_autoregression)
       layer = layer_class(net, labels_one_hot, self._params, mparams)
@@ -252,16 +389,16 @@ class Model(object):
       A tensor with the same size as any input tensors.
     """
     batch_size, height, width, num_features = [
-      d.value for d in nets_list[0].get_shape().dims
+        d.value for d in nets_list[0].get_shape().dims
     ]
     xy_flat_shape = (batch_size, 1, height * width, num_features)
     nets_for_merge = []
-    with tf.variable_scope('max_pool_views', values=nets_list):
+    with tf.compat.v1.variable_scope('max_pool_views', values=nets_list):
       for net in nets_list:
         nets_for_merge.append(tf.reshape(net, xy_flat_shape))
       merged_net = tf.concat(nets_for_merge, 1)
       net = slim.max_pool2d(
-        merged_net, kernel_size=[len(nets_list), 1], stride=1)
+          merged_net, kernel_size=[len(nets_list), 1], stride=1)
       net = tf.reshape(net, (batch_size, height, width, num_features))
     return net
 
@@ -277,18 +414,20 @@ class Model(object):
     Returns:
       A tensor of shape [batch_size, seq_length, features_size].
     """
-    with tf.variable_scope('pool_views_fn/STCK'):
+    with tf.compat.v1.variable_scope('pool_views_fn/STCK'):
       net = tf.concat(nets, 1)
-      batch_size = net.get_shape().dims[0].value
+      batch_size = tf.shape(input=net)[0]
+      image_size = net.get_shape().dims[1].value * \
+          net.get_shape().dims[2].value
       feature_size = net.get_shape().dims[3].value
-      return tf.reshape(net, [batch_size, -1, feature_size])
+      return tf.reshape(net, tf.stack([batch_size, image_size, feature_size]))
 
   def char_predictions(self, chars_logit):
     """Returns confidence scores (softmax values) for predicted characters.
 
     Args:
-      chars_logit: chars logits, a tensor with shape
-        [batch_size x seq_length x num_char_classes]
+      chars_logit: chars logits, a tensor with shape [batch_size x seq_length x
+        num_char_classes]
 
     Returns:
       A tuple (ids, log_prob, scores), where:
@@ -301,12 +440,17 @@ class Model(object):
           with shape [batch_size x seq_length].
     """
     log_prob = utils.logits_to_log_prob(chars_logit)
-    ids = tf.to_int32(tf.argmax(log_prob, axis=2), name='predicted_chars')
+    ids = tf.cast(tf.argmax(input=log_prob, axis=2),
+                  name='predicted_chars', dtype=tf.int32)
     mask = tf.cast(
-      slim.one_hot_encoding(ids, self._params.num_char_classes), tf.bool)
+        slim.one_hot_encoding(ids, self._params.num_char_classes), tf.bool)
     all_scores = tf.nn.softmax(chars_logit)
-    selected_scores = tf.boolean_mask(all_scores, mask, name='char_scores')
-    scores = tf.reshape(selected_scores, shape=(-1, self._params.seq_length))
+    selected_scores = tf.boolean_mask(
+        tensor=all_scores, mask=mask, name='char_scores')
+    scores = tf.reshape(
+        selected_scores,
+        shape=(-1, self._params.seq_length),
+        name='predicted_scores')
     return ids, log_prob, scores
 
   def encode_coordinates_fn(self, net):
@@ -323,12 +467,12 @@ class Model(object):
     """
     mparams = self._mparams['encode_coordinates_fn']
     if mparams.enabled:
-      batch_size, h, w, _ = net.shape.as_list()
+      batch_size, h, w, _ = get_tensor_dimensions(net)
       x, y = tf.meshgrid(tf.range(w), tf.range(h))
       w_loc = slim.one_hot_encoding(x, num_classes=w)
       h_loc = slim.one_hot_encoding(y, num_classes=h)
       loc = tf.concat([h_loc, w_loc], 2)
-      loc = tf.tile(tf.expand_dims(loc, 0), [batch_size, 1, 1, 1])
+      loc = tf.tile(tf.expand_dims(loc, 0), tf.stack([batch_size, 1, 1, 1]))
       return tf.concat([net, loc], 3)
     else:
       return net
@@ -341,7 +485,8 @@ class Model(object):
     """Creates a base part of the Model (no gradients, losses or summaries).
 
     Args:
-      images: A tensor of shape [batch_size, height, width, channels].
+      images: A tensor of shape [batch_size, height, width, channels] with pixel
+        values in the range [0.0, 1.0].
       labels_one_hot: Optional (can be None) one-hot encoding for ground truth
         labels. If provided the function will create a model for training.
       scope: Optional variable_scope.
@@ -353,14 +498,19 @@ class Model(object):
     """
     logging.debug('images: %s', images)
     is_training = labels_one_hot is not None
-    with tf.variable_scope(scope, reuse=reuse):
+
+    # Normalize image pixel values to have a symmetrical range around zero.
+    images = tf.subtract(images, 0.5)
+    images = tf.multiply(images, 2.5)
+
+    with tf.compat.v1.variable_scope(scope, reuse=reuse):
       views = tf.split(
-        value=images, num_or_size_splits=self._params.num_views, axis=2)
+          value=images, num_or_size_splits=self._params.num_views, axis=2)
       logging.debug('Views=%d single view: %s', len(views), views[0])
 
       nets = [
-        self.conv_tower_fn(v, is_training, reuse=(i != 0))
-        for i, v in enumerate(views)
+          self.conv_tower_fn(v, is_training, reuse=(i != 0))
+          for i, v in enumerate(views)
       ]
       logging.debug('Conv tower: %s', nets[0])
 
@@ -374,18 +524,34 @@ class Model(object):
       logging.debug('chars_logit: %s', chars_logit)
 
       predicted_chars, chars_log_prob, predicted_scores = (
-        self.char_predictions(chars_logit))
+          self.char_predictions(chars_logit))
       if self._charset:
         character_mapper = CharsetMapper(self._charset)
         predicted_text = character_mapper.get_text(predicted_chars)
       else:
         predicted_text = tf.constant([])
+
+      text_log_prob, predicted_length = null_based_length_prediction(
+          chars_log_prob, self._params.null_code)
+      predicted_conf = lookup_indexed_value(predicted_length, text_log_prob)
+      # Convert predicted confidence from sum of logs to geometric mean
+      normalized_seq_conf = tf.exp(
+          tf.divide(predicted_conf,
+                    tf.cast(predicted_length + 1, predicted_conf.dtype)),
+          name='normalized_seq_conf')
+      predicted_conf = tf.identity(predicted_conf, name='predicted_conf')
+      predicted_text = tf.identity(predicted_text, name='predicted_text')
+      predicted_length = tf.identity(predicted_length, name='predicted_length')
+
     return OutputEndpoints(
-      chars_logit=chars_logit,
-      chars_log_prob=chars_log_prob,
-      predicted_chars=predicted_chars,
-      predicted_scores=predicted_scores,
-      predicted_text=predicted_text)
+        chars_logit=chars_logit,
+        chars_log_prob=chars_log_prob,
+        predicted_chars=predicted_chars,
+        predicted_scores=predicted_scores,
+        predicted_length=predicted_length,
+        predicted_text=predicted_text,
+        predicted_conf=predicted_conf,
+        normalized_seq_conf=normalized_seq_conf)
 
   def create_loss(self, data, endpoints):
     """Creates all losses required to train the model.
@@ -404,7 +570,7 @@ class Model(object):
     # multiple losses including regularization losses.
     self.sequence_loss_fn(endpoints.chars_logit, data.labels)
     total_loss = slim.losses.get_total_loss()
-    tf.summary.scalar('TotalLoss', total_loss)
+    tf.compat.v1.summary.scalar('TotalLoss', total_loss)
     return total_loss
 
   def label_smoothing_regularization(self, chars_labels, weight=0.1):
@@ -413,15 +579,15 @@ class Model(object):
     Uses the same method as in https://arxiv.org/abs/1512.00567.
 
     Args:
-      chars_labels: ground truth ids of charactes,
-        shape=[batch_size, seq_length];
+      chars_labels: ground truth ids of charactes, shape=[batch_size,
+        seq_length];
       weight: label-smoothing regularization weight.
 
     Returns:
       A sensor with the same shape as the input.
     """
     one_hot_labels = tf.one_hot(
-      chars_labels, depth=self._params.num_char_classes, axis=-1)
+        chars_labels, depth=self._params.num_char_classes, axis=-1)
     pos_weight = 1.0 - weight
     neg_weight = weight / self._params.num_char_classes
     return one_hot_labels * pos_weight + neg_weight
@@ -433,20 +599,20 @@ class Model(object):
     also ignore all null chars after the first one.
 
     Args:
-      chars_logits: logits for predicted characters,
-        shape=[batch_size, seq_length, num_char_classes];
-      chars_labels: ground truth ids of characters,
-        shape=[batch_size, seq_length];
+      chars_logits: logits for predicted characters, shape=[batch_size,
+        seq_length, num_char_classes];
+      chars_labels: ground truth ids of characters, shape=[batch_size,
+        seq_length];
       mparams: method hyper parameters.
 
     Returns:
       A Tensor with shape [batch_size] - the log-perplexity for each sequence.
     """
     mparams = self._mparams['sequence_loss_fn']
-    with tf.variable_scope('sequence_loss_fn/SLF'):
+    with tf.compat.v1.variable_scope('sequence_loss_fn/SLF'):
       if mparams.label_smoothing > 0:
         smoothed_one_hot_labels = self.label_smoothing_regularization(
-          chars_labels, mparams.label_smoothing)
+            chars_labels, mparams.label_smoothing)
         labels_list = tf.unstack(smoothed_one_hot_labels, axis=1)
       else:
         # NOTE: in case of sparse softmax we are not using one-hot
@@ -459,21 +625,21 @@ class Model(object):
       else:
         # Suppose that reject character is the last in the charset.
         reject_char = tf.constant(
-          self._params.num_char_classes - 1,
-          shape=(batch_size, seq_length),
-          dtype=tf.int64)
+            self._params.num_char_classes - 1,
+            shape=(batch_size, seq_length),
+            dtype=tf.int64)
         known_char = tf.not_equal(chars_labels, reject_char)
-        weights = tf.to_float(known_char)
+        weights = tf.cast(known_char, dtype=tf.float32)
 
       logits_list = tf.unstack(chars_logits, axis=1)
       weights_list = tf.unstack(weights, axis=1)
       loss = tf.contrib.legacy_seq2seq.sequence_loss(
-        logits_list,
-        labels_list,
-        weights_list,
-        softmax_loss_function=get_softmax_loss_fn(mparams.label_smoothing),
-        average_across_timesteps=mparams.average_across_timesteps)
-      tf.losses.add_loss(loss)
+          logits_list,
+          labels_list,
+          weights_list,
+          softmax_loss_function=get_softmax_loss_fn(mparams.label_smoothing),
+          average_across_timesteps=mparams.average_across_timesteps)
+      tf.compat.v1.losses.add_loss(loss)
       return loss
 
   def create_summaries(self, data, endpoints, charset, is_training):
@@ -482,8 +648,8 @@ class Model(object):
     Args:
       data: InputEndpoints namedtuple.
       endpoints: OutputEndpoints namedtuple.
-      charset: A dictionary with mapping between character codes and
-        unicode characters. Use the one provided by a dataset.charset.
+      charset: A dictionary with mapping between character codes and unicode
+        characters. Use the one provided by a dataset.charset.
       is_training: If True will create summary prefixes for training job,
         otherwise - for evaluation.
 
@@ -503,13 +669,14 @@ class Model(object):
     # tf.summary.text(sname('text/pr'), pr_text)
     # gt_text = charset_mapper.get_text(data.labels[:max_outputs,:])
     # tf.summary.text(sname('text/gt'), gt_text)
-    tf.summary.image(sname('image'), data.images, max_outputs=max_outputs)
+    tf.compat.v1.summary.image(
+        sname('image'), data.images, max_outputs=max_outputs)
 
     if is_training:
-      tf.summary.image(
-        sname('image/orig'), data.images_orig, max_outputs=max_outputs)
-      for var in tf.trainable_variables():
-        tf.summary.histogram(var.op.name, var)
+      tf.compat.v1.summary.image(
+          sname('image/orig'), data.images_orig, max_outputs=max_outputs)
+      for var in tf.compat.v1.trainable_variables():
+        tf.compat.v1.summary.histogram(var.op.name, var)
       return None
 
     else:
@@ -520,32 +687,36 @@ class Model(object):
         names_to_values[name] = value_update_tuple[0]
         names_to_updates[name] = value_update_tuple[1]
 
-      use_metric('CharacterAccuracy',
-                 metrics.char_accuracy(
-                   endpoints.predicted_chars,
-                   data.labels,
-                   streaming=True,
-                   rej_char=self._params.null_code))
+      use_metric(
+          'CharacterAccuracy',
+          metrics.char_accuracy(
+              endpoints.predicted_chars,
+              data.labels,
+              streaming=True,
+              rej_char=self._params.null_code))
       # Sequence accuracy computed by cutting sequence at the first null char
-      use_metric('SequenceAccuracy',
-                 metrics.sequence_accuracy(
-                   endpoints.predicted_chars,
-                   data.labels,
-                   streaming=True,
-                   rej_char=self._params.null_code))
+      use_metric(
+          'SequenceAccuracy',
+          metrics.sequence_accuracy(
+              endpoints.predicted_chars,
+              data.labels,
+              streaming=True,
+              rej_char=self._params.null_code))
 
       for name, value in names_to_values.items():
         summary_name = 'eval/' + name
-        tf.summary.scalar(summary_name, tf.Print(value, [value], summary_name))
+        tf.compat.v1.summary.scalar(
+            summary_name, tf.compat.v1.Print(value, [value], summary_name))
       return list(names_to_updates.values())
 
-  def create_init_fn_to_restore(self, master_checkpoint,
+  def create_init_fn_to_restore(self,
+                                master_checkpoint,
                                 inception_checkpoint=None):
     """Creates an init operations to restore weights from various checkpoints.
 
     Args:
-      master_checkpoint: path to a checkpoint which contains all weights for
-        the whole model.
+      master_checkpoint: path to a checkpoint which contains all weights for the
+        whole model.
       inception_checkpoint: path to a checkpoint which contains weights for the
         inception part only.
 
@@ -556,8 +727,8 @@ class Model(object):
     all_feed_dict = {}
 
     def assign_from_checkpoint(variables, checkpoint):
-      logging.info('Request to re-store %d weights from %s',
-                   len(variables), checkpoint)
+      logging.info('Request to re-store %d weights from %s', len(variables),
+                   checkpoint)
       if not variables:
         logging.error('Can\'t find any variables to restore.')
         sys.exit(1)
@@ -565,15 +736,18 @@ class Model(object):
       all_assign_ops.append(assign_op)
       all_feed_dict.update(feed_dict)
 
-    logging.info('variables_to_restore:\n%s' % utils.variables_to_restore().keys())
-    logging.info('moving_average_variables:\n%s' % [v.op.name for v in tf.moving_average_variables()])
-    logging.info('trainable_variables:\n%s' % [v.op.name for v in tf.trainable_variables()])
+    logging.info('variables_to_restore:\n%s',
+                 utils.variables_to_restore().keys())
+    logging.info('moving_average_variables:\n%s',
+                 [v.op.name for v in tf.compat.v1.moving_average_variables()])
+    logging.info('trainable_variables:\n%s',
+                 [v.op.name for v in tf.compat.v1.trainable_variables()])
     if master_checkpoint:
       assign_from_checkpoint(utils.variables_to_restore(), master_checkpoint)
 
     if inception_checkpoint:
       variables = utils.variables_to_restore(
-        'AttentionOcr_v1/conv_tower_fn/INCE', strip_scope=True)
+          'AttentionOcr_v1/conv_tower_fn/INCE', strip_scope=True)
       assign_from_checkpoint(variables, inception_checkpoint)
 
     def init_assign_fn(sess):

research/attention_ocr/python/model_export.py

+# Lint as: python3
+# Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ==============================================================================
+r"""Converts existing checkpoint into a SavedModel.
+
+Usage example:
+python model_export.py \
+  --logtostderr --checkpoint=model.ckpt-399731 \
+  --export_dir=/tmp/attention_ocr_export
+"""
+import os
+
+import tensorflow as tf
+from tensorflow import app
+from tensorflow.contrib import slim
+from tensorflow.python.platform import flags
+
+import common_flags
+import model_export_lib
+
+FLAGS = flags.FLAGS
+common_flags.define()
+
+flags.DEFINE_string('export_dir', None, 'Directory to export model files to.')
+flags.DEFINE_integer(
+    'image_width', None,
+    'Image width used during training (or crop width if used)'
+    ' If not set, the dataset default is used instead.')
+flags.DEFINE_integer(
+    'image_height', None,
+    'Image height used during training(or crop height if used)'
+    ' If not set, the dataset default is used instead.')
+flags.DEFINE_string('work_dir', '/tmp',
+                    'A directory to store temporary files.')
+flags.DEFINE_integer('version_number', 1, 'Version number of the model')
+flags.DEFINE_bool(
+    'export_for_serving', True,
+    'Whether the exported model accepts serialized tf.Example '
+    'protos as input')
+
+
+def get_checkpoint_path():
+  """Returns a path to a checkpoint based on specified commandline flags.
+
+  In order to specify a full path to a checkpoint use --checkpoint flag.
+  Alternatively, if --train_log_dir was specified it will return a path to the
+  most recent checkpoint.
+
+  Raises:
+    ValueError: in case it can't find a checkpoint.
+
+  Returns:
+    A string.
+  """
+  if FLAGS.checkpoint:
+    return FLAGS.checkpoint
+  else:
+    model_save_path = tf.train.latest_checkpoint(FLAGS.train_log_dir)
+    if not model_save_path:
+      raise ValueError('Can\'t find a checkpoint in: %s' % FLAGS.train_log_dir)
+    return model_save_path
+
+
+def export_model(export_dir,
+                 export_for_serving,
+                 batch_size=None,
+                 crop_image_width=None,
+                 crop_image_height=None):
+  """Exports a model to the named directory.
+
+  Note that --datatset_name and --checkpoint are required and parsed by the
+  underlying module common_flags.
+
+  Args:
+    export_dir: The output dir where model is exported to.
+    export_for_serving: If True, expects a serialized image as input and attach
+      image normalization as part of exported graph.
+    batch_size: For non-serving export, the input batch_size needs to be
+      specified.
+    crop_image_width: Width of the input image. Uses the dataset default if
+      None.
+    crop_image_height: Height of the input image. Uses the dataset default if
+      None.
+
+  Returns:
+    Returns the model signature_def.
+  """
+  # Dataset object used only to get all parameters for the model.
+  dataset = common_flags.create_dataset(split_name='test')
+  model = common_flags.create_model(
+      dataset.num_char_classes,
+      dataset.max_sequence_length,
+      dataset.num_of_views,
+      dataset.null_code,
+      charset=dataset.charset)
+  dataset_image_height, dataset_image_width, image_depth = dataset.image_shape
+
+  # Add check for charmap file
+  if not os.path.exists(dataset.charset_file):
+    raise ValueError('No charset defined at {}: export will fail'.format(
+        dataset.charset))
+
+  # Default to dataset dimensions, otherwise use provided dimensions.
+  image_width = crop_image_width or dataset_image_width
+  image_height = crop_image_height or dataset_image_height
+
+  if export_for_serving:
+    images_orig = tf.compat.v1.placeholder(
+        tf.string, shape=[batch_size], name='tf_example')
+    images_orig_float = model_export_lib.generate_tfexample_image(
+        images_orig,
+        image_height,
+        image_width,
+        image_depth,
+        name='float_images')
+  else:
+    images_shape = (batch_size, image_height, image_width, image_depth)
+    images_orig = tf.compat.v1.placeholder(
+        tf.uint8, shape=images_shape, name='original_image')
+    images_orig_float = tf.image.convert_image_dtype(
+        images_orig, dtype=tf.float32, name='float_images')
+
+  endpoints = model.create_base(images_orig_float, labels_one_hot=None)
+
+  sess = tf.compat.v1.Session()
+  saver = tf.compat.v1.train.Saver(
+      slim.get_variables_to_restore(), sharded=True)
+  saver.restore(sess, get_checkpoint_path())
+  tf.compat.v1.logging.info('Model restored successfully.')
+
+  # Create model signature.
+  if export_for_serving:
+    input_tensors = {
+        tf.saved_model.CLASSIFY_INPUTS: images_orig
+    }
+  else:
+    input_tensors = {'images': images_orig}
+  signature_inputs = model_export_lib.build_tensor_info(input_tensors)
+  # NOTE: Tensors 'image_float' and 'chars_logit' are used by the inference
+  # or to compute saliency maps.
+  output_tensors = {
+      'images_float': images_orig_float,
+      'predictions': endpoints.predicted_chars,
+      'scores': endpoints.predicted_scores,
+      'chars_logit': endpoints.chars_logit,
+      'predicted_length': endpoints.predicted_length,
+      'predicted_text': endpoints.predicted_text,
+      'predicted_conf': endpoints.predicted_conf,
+      'normalized_seq_conf': endpoints.normalized_seq_conf
+  }
+  for i, t in enumerate(
+      model_export_lib.attention_ocr_attention_masks(
+          dataset.max_sequence_length)):
+    output_tensors['attention_mask_%d' % i] = t
+  signature_outputs = model_export_lib.build_tensor_info(output_tensors)
+  signature_def = tf.compat.v1.saved_model.signature_def_utils.build_signature_def(
+      signature_inputs, signature_outputs,
+      tf.saved_model.CLASSIFY_METHOD_NAME)
+  # Save model.
+  builder = tf.compat.v1.saved_model.builder.SavedModelBuilder(export_dir)
+  builder.add_meta_graph_and_variables(
+      sess, [tf.saved_model.SERVING],
+      signature_def_map={
+          tf.saved_model.DEFAULT_SERVING_SIGNATURE_DEF_KEY:
+              signature_def
+      },
+      main_op=tf.compat.v1.tables_initializer(),
+      strip_default_attrs=True)
+  builder.save()
+  tf.compat.v1.logging.info('Model has been exported to %s' % export_dir)
+
+  return signature_def
+
+
+def main(unused_argv):
+  if os.path.exists(FLAGS.export_dir):
+    raise ValueError('export_dir already exists: exporting will fail')
+
+  export_model(FLAGS.export_dir, FLAGS.export_for_serving, FLAGS.batch_size,
+               FLAGS.image_width, FLAGS.image_height)
+
+
+if __name__ == '__main__':
+  flags.mark_flag_as_required('dataset_name')
+  flags.mark_flag_as_required('export_dir')
+  app.run(main)

research/attention_ocr/python/model_export_lib.py

+# Lint as: python3
+# Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ==============================================================================
+"""Utility functions for exporting Attention OCR model."""
+
+import tensorflow as tf
+
+
+# Function borrowed from research/object_detection/core/preprocessor.py
+def normalize_image(image, original_minval, original_maxval, target_minval,
+                    target_maxval):
+  """Normalizes pixel values in the image.
+
+  Moves the pixel values from the current [original_minval, original_maxval]
+  range to a the [target_minval, target_maxval] range.
+  Args:
+    image: rank 3 float32 tensor containing 1 image -> [height, width,
+      channels].
+    original_minval: current image minimum value.
+    original_maxval: current image maximum value.
+    target_minval: target image minimum value.
+    target_maxval: target image maximum value.
+
+  Returns:
+    image: image which is the same shape as input image.
+  """
+  with tf.compat.v1.name_scope('NormalizeImage', values=[image]):
+    original_minval = float(original_minval)
+    original_maxval = float(original_maxval)
+    target_minval = float(target_minval)
+    target_maxval = float(target_maxval)
+    image = tf.cast(image, dtype=tf.float32)
+    image = tf.subtract(image, original_minval)
+    image = tf.multiply(image, (target_maxval - target_minval) /
+                        (original_maxval - original_minval))
+    image = tf.add(image, target_minval)
+    return image
+
+
+def generate_tfexample_image(input_example_strings,
+                             image_height,
+                             image_width,
+                             image_channels,
+                             name=None):
+  """Parses a 1D tensor of serialized tf.Example protos and returns image batch.
+
+  Args:
+    input_example_strings: A 1-Dimensional tensor of size [batch_size] and type
+      tf.string containing a serialized Example proto per image.
+    image_height: First image dimension.
+    image_width: Second image dimension.
+    image_channels: Third image dimension.
+    name: optional tensor name.
+
+  Returns:
+    A tensor with shape [batch_size, height, width, channels] of type float32
+    with values in the range [0..1]
+  """
+  batch_size = tf.shape(input=input_example_strings)[0]
+  images_shape = tf.stack(
+      [batch_size, image_height, image_width, image_channels])
+  tf_example_image_key = 'image/encoded'
+  feature_configs = {
+      tf_example_image_key:
+          tf.io.FixedLenFeature(
+              image_height * image_width * image_channels, dtype=tf.float32)
+  }
+  feature_tensors = tf.io.parse_example(
+      serialized=input_example_strings, features=feature_configs)
+  float_images = tf.reshape(
+      normalize_image(
+          feature_tensors[tf_example_image_key],
+          original_minval=0.0,
+          original_maxval=255.0,
+          target_minval=0.0,
+          target_maxval=1.0),
+      images_shape,
+      name=name)
+  return float_images
+
+
+def attention_ocr_attention_masks(num_characters):
+  # TODO(gorban): use tensors directly after replacing LSTM unroll methods.
+  prefix = ('AttentionOcr_v1/'
+            'sequence_logit_fn/SQLR/LSTM/attention_decoder/Attention_0')
+  names = ['%s/Softmax:0' % (prefix)]
+  for i in range(1, num_characters):
+    names += ['%s_%d/Softmax:0' % (prefix, i)]
+  return [tf.compat.v1.get_default_graph().get_tensor_by_name(n) for n in names]
+
+
+def build_tensor_info(tensor_dict):
+  return {
+      k: tf.compat.v1.saved_model.utils.build_tensor_info(t)
+      for k, t in tensor_dict.items()
+  }

research/attention_ocr/python/model_export_test.py

+# Lint as: python3
+# Copyright 2020 The TensorFlow Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ==============================================================================
+"""Tests for model_export."""
+import os
+
+import numpy as np
+from absl.testing import flagsaver
+import tensorflow as tf
+
+import common_flags
+import model_export
+
+_CHECKPOINT = 'model.ckpt-399731'
+_CHECKPOINT_URL = (
+    'http://download.tensorflow.org/models/attention_ocr_2017_08_09.tar.gz')
+
+
+def _clean_up():
+  tf.io.gfile.rmtree(tf.compat.v1.test.get_temp_dir())
+
+
+def _create_tf_example_string(image):
+  """Create a serialized tf.Example proto for feeding the model."""
+  example = tf.train.Example()
+  example.features.feature['image/encoded'].float_list.value.extend(
+      list(np.reshape(image, (-1))))
+  return example.SerializeToString()
+
+
+class AttentionOcrExportTest(tf.test.TestCase):
+  """Tests for model_export.export_model."""
+
+  def setUp(self):
+    for suffix in ['.meta', '.index', '.data-00000-of-00001']:
+      filename = _CHECKPOINT + suffix
+      self.assertTrue(
+          tf.io.gfile.exists(filename),
+          msg='Missing checkpoint file %s. '
+          'Please download and extract it from %s' %
+          (filename, _CHECKPOINT_URL))
+    tf.flags.FLAGS.dataset_name = 'fsns'
+    tf.flags.FLAGS.checkpoint = _CHECKPOINT
+    tf.flags.FLAGS.dataset_dir = os.path.join(
+        os.path.dirname(__file__), 'datasets/testdata/fsns')
+    tf.test.TestCase.setUp(self)
+    _clean_up()
+    self.export_dir = os.path.join(
+        tf.compat.v1.test.get_temp_dir(), 'exported_model')
+    self.minimal_output_signature = {
+        'predictions': 'AttentionOcr_v1/predicted_chars:0',
+        'scores': 'AttentionOcr_v1/predicted_scores:0',
+        'predicted_length': 'AttentionOcr_v1/predicted_length:0',
+        'predicted_text': 'AttentionOcr_v1/predicted_text:0',
+        'predicted_conf': 'AttentionOcr_v1/predicted_conf:0',
+        'normalized_seq_conf': 'AttentionOcr_v1/normalized_seq_conf:0'
+    }
+
+  def create_input_feed(self, graph_def, serving):
+    """Returns the input feed for the model.
+
+    Creates random images, according to the size specified by dataset_name,
+    format it in the correct way depending on whether the model was exported
+    for serving, and return the correctly keyed feed_dict for inference.
+
+    Args:
+      graph_def: Graph definition of the loaded model.
+      serving: Whether the model was exported for Serving.
+
+    Returns:
+      The feed_dict suitable for model inference.
+    """
+    # Creates a dataset based on FLAGS.dataset_name.
+    self.dataset = common_flags.create_dataset('test')
+    # Create some random images to test inference for any dataset.
+    self.images = {
+        'img1':
+            np.random.uniform(low=64, high=192,
+                              size=self.dataset.image_shape).astype('uint8'),
+        'img2':
+            np.random.uniform(low=32, high=224,
+                              size=self.dataset.image_shape).astype('uint8'),
+    }
+    signature_def = graph_def.signature_def[
+        tf.saved_model.DEFAULT_SERVING_SIGNATURE_DEF_KEY]
+    if serving:
+      input_name = signature_def.inputs[
+          tf.saved_model.CLASSIFY_INPUTS].name
+      # Model for serving takes input: inputs['inputs'] = 'tf_example:0'
+      feed_dict = {
+          input_name: [
+              _create_tf_example_string(self.images['img1']),
+              _create_tf_example_string(self.images['img2'])
+          ]
+      }
+    else:
+      input_name = signature_def.inputs['images'].name
+      # Model for direct use takes input: inputs['images'] = 'original_image:0'
+      feed_dict = {
+          input_name: np.stack([self.images['img1'], self.images['img2']])
+      }
+    return feed_dict
+
+  def verify_export_load_and_inference(self, export_for_serving=False):
+    """Verify exported model can be loaded and inference can run successfully.
+
+    This function will load the exported model in self.export_dir, then create
+    some fake images according to the specification of FLAGS.dataset_name.
+    It then feeds the input through the model, and verify the minimal set of
+    output signatures are present.
+    Note: Model and dataset creation in the underlying library depends on the
+          following commandline flags:
+            FLAGS.dataset_name
+    Args:
+      export_for_serving: True if the model was exported for Serving. This
+        affects how input is fed into the model.
+    """
+    tf.compat.v1.reset_default_graph()
+    sess = tf.compat.v1.Session()
+    graph_def = tf.compat.v1.saved_model.loader.load(
+        sess=sess,
+        tags=[tf.saved_model.SERVING],
+        export_dir=self.export_dir)
+    feed_dict = self.create_input_feed(graph_def, export_for_serving)
+    results = sess.run(self.minimal_output_signature, feed_dict=feed_dict)
+
+    out_shape = (2,)
+    self.assertEqual(np.shape(results['predicted_conf']), out_shape)
+    self.assertEqual(np.shape(results['predicted_text']), out_shape)
+    self.assertEqual(np.shape(results['predicted_length']), out_shape)
+    self.assertEqual(np.shape(results['normalized_seq_conf']), out_shape)
+    out_shape = (2, self.dataset.max_sequence_length)
+    self.assertEqual(np.shape(results['scores']), out_shape)
+    self.assertEqual(np.shape(results['predictions']), out_shape)
+
+  @flagsaver.flagsaver
+  def test_fsns_export_for_serving_and_load_inference(self):
+    model_export.export_model(self.export_dir, True)
+    self.verify_export_load_and_inference(True)
+
+  @flagsaver.flagsaver
+  def test_fsns_export_and_load_inference(self):
+    model_export.export_model(self.export_dir, False, batch_size=2)
+    self.verify_export_load_and_inference(False)
+
+
+if __name__ == '__main__':
+  tf.test.main()

research/attention_ocr/python/model_test.py

@@ -12,11 +12,10 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 # ==============================================================================
-
 """Tests for the model."""
+import string
 
 import numpy as np
-import string
 import tensorflow as tf
 from tensorflow.contrib import slim
 
@@ -32,6 +31,7 @@ def create_fake_charset(num_char_classes):
 
 
 class ModelTest(tf.test.TestCase):
+
   def setUp(self):
     tf.test.TestCase.setUp(self)
 
@@ -51,18 +51,21 @@ class ModelTest(tf.test.TestCase):
     self.chars_logit_shape = (self.batch_size, self.seq_length,
                               self.num_char_classes)
     self.length_logit_shape = (self.batch_size, self.seq_length + 1)
+    # Placeholder knows image dimensions, but not batch size.
+    self.input_images = tf.compat.v1.placeholder(
+        tf.float32,
+        shape=(None, self.image_height, self.image_width, 3),
+        name='input_node')
 
     self.initialize_fakes()
 
   def initialize_fakes(self):
     self.images_shape = (self.batch_size, self.image_height, self.image_width,
                          3)
-    self.fake_images = tf.constant(
-        self.rng.randint(low=0, high=255,
-                         size=self.images_shape).astype('float32'),
-        name='input_node')
-    self.fake_conv_tower_np = self.rng.randn(
-        *self.conv_tower_shape).astype('float32')
+    self.fake_images = self.rng.randint(
+        low=0, high=255, size=self.images_shape).astype('float32')
+    self.fake_conv_tower_np = self.rng.randn(*self.conv_tower_shape).astype(
+        'float32')
     self.fake_conv_tower = tf.constant(self.fake_conv_tower_np)
     self.fake_logits = tf.constant(
         self.rng.randn(*self.chars_logit_shape).astype('float32'))
@@ -74,33 +77,44 @@ class ModelTest(tf.test.TestCase):
 
   def create_model(self, charset=None):
     return model.Model(
-        self.num_char_classes, self.seq_length, num_views=4, null_code=62,
+        self.num_char_classes,
+        self.seq_length,
+        num_views=4,
+        null_code=62,
         charset=charset)
 
   def test_char_related_shapes(self):
-    ocr_model = self.create_model()
+    charset = create_fake_charset(self.num_char_classes)
+    ocr_model = self.create_model(charset=charset)
     with self.test_session() as sess:
       endpoints_tf = ocr_model.create_base(
-          images=self.fake_images, labels_one_hot=None)
-
-      sess.run(tf.global_variables_initializer())
-      endpoints = sess.run(endpoints_tf)
-
-      self.assertEqual((self.batch_size, self.seq_length,
-                        self.num_char_classes), endpoints.chars_logit.shape)
-      self.assertEqual((self.batch_size, self.seq_length,
-                        self.num_char_classes), endpoints.chars_log_prob.shape)
+          images=self.input_images, labels_one_hot=None)
+      sess.run(tf.compat.v1.global_variables_initializer())
+      tf.compat.v1.tables_initializer().run()
+      endpoints = sess.run(
+          endpoints_tf, feed_dict={self.input_images: self.fake_images})
+
+      self.assertEqual(
+          (self.batch_size, self.seq_length, self.num_char_classes),
+          endpoints.chars_logit.shape)
+      self.assertEqual(
+          (self.batch_size, self.seq_length, self.num_char_classes),
+          endpoints.chars_log_prob.shape)
       self.assertEqual((self.batch_size, self.seq_length),
                        endpoints.predicted_chars.shape)
       self.assertEqual((self.batch_size, self.seq_length),
                        endpoints.predicted_scores.shape)
+      self.assertEqual((self.batch_size,), endpoints.predicted_text.shape)
+      self.assertEqual((self.batch_size,), endpoints.predicted_conf.shape)
+      self.assertEqual((self.batch_size,), endpoints.normalized_seq_conf.shape)
 
   def test_predicted_scores_are_within_range(self):
     ocr_model = self.create_model()
 
     _, _, scores = ocr_model.char_predictions(self.fake_logits)
     with self.test_session() as sess:
-      scores_np = sess.run(scores)
+      scores_np = sess.run(
+          scores, feed_dict={self.input_images: self.fake_images})
 
     values_in_range = (scores_np >= 0.0) & (scores_np <= 1.0)
     self.assertTrue(
@@ -111,10 +125,11 @@ class ModelTest(tf.test.TestCase):
   def test_conv_tower_shape(self):
     with self.test_session() as sess:
       ocr_model = self.create_model()
-      conv_tower = ocr_model.conv_tower_fn(self.fake_images)
+      conv_tower = ocr_model.conv_tower_fn(self.input_images)
 
-      sess.run(tf.global_variables_initializer())
-      conv_tower_np = sess.run(conv_tower)
+      sess.run(tf.compat.v1.global_variables_initializer())
+      conv_tower_np = sess.run(
+          conv_tower, feed_dict={self.input_images: self.fake_images})
 
       self.assertEqual(self.conv_tower_shape, conv_tower_np.shape)
 
@@ -124,11 +139,12 @@ class ModelTest(tf.test.TestCase):
     # updates, gradients and variances. It also depends on the type of used
     # optimizer.
     ocr_model = self.create_model()
-    ocr_model.create_base(images=self.fake_images, labels_one_hot=None)
+    ocr_model.create_base(images=self.input_images, labels_one_hot=None)
     with self.test_session() as sess:
-      tfprof_root = tf.profiler.profile(
+      tfprof_root = tf.compat.v1.profiler.profile(
           sess.graph,
-          options=tf.profiler.ProfileOptionBuilder.trainable_variables_parameter())
+          options=tf.compat.v1.profiler.ProfileOptionBuilder
+          .trainable_variables_parameter())
 
       model_size_bytes = 4 * tfprof_root.total_parameters
       self.assertLess(model_size_bytes, 1 * 2**30)
@@ -147,9 +163,9 @@ class ModelTest(tf.test.TestCase):
     summaries = ocr_model.create_summaries(
         data, endpoints, charset, is_training=False)
     with self.test_session() as sess:
-      sess.run(tf.global_variables_initializer())
-      sess.run(tf.local_variables_initializer())
-      tf.tables_initializer().run()
+      sess.run(tf.compat.v1.global_variables_initializer())
+      sess.run(tf.compat.v1.local_variables_initializer())
+      tf.compat.v1.tables_initializer().run()
       sess.run(summaries)  # just check it is runnable
 
   def test_sequence_loss_function_without_label_smoothing(self):
@@ -158,7 +174,7 @@ class ModelTest(tf.test.TestCase):
 
     loss = model.sequence_loss_fn(self.fake_logits, self.fake_labels)
     with self.test_session() as sess:
-      loss_np = sess.run(loss)
+      loss_np = sess.run(loss, feed_dict={self.input_images: self.fake_images})
 
     # This test checks that the loss function is 'runnable'.
     self.assertEqual(loss_np.shape, tuple())
@@ -172,19 +188,21 @@ class ModelTest(tf.test.TestCase):
     Returns:
       a list of tensors with encoded image coordinates in them.
     """
-    batch_size, h, w, _ = net.shape.as_list()
+    batch_size = tf.shape(input=net)[0]
+    _, h, w, _ = net.shape.as_list()
     h_loc = [
-      tf.tile(
-          tf.reshape(
-              tf.contrib.layers.one_hot_encoding(
-                  tf.constant([i]), num_classes=h), [h, 1]), [1, w])
-      for i in range(h)
+        tf.tile(
+            tf.reshape(
+                tf.contrib.layers.one_hot_encoding(
+                    tf.constant([i]), num_classes=h), [h, 1]), [1, w])
+        for i in range(h)
     ]
     h_loc = tf.concat([tf.expand_dims(t, 2) for t in h_loc], 2)
     w_loc = [
-      tf.tile(
-          tf.contrib.layers.one_hot_encoding(tf.constant([i]), num_classes=w),
-          [h, 1]) for i in range(w)
+        tf.tile(
+            tf.contrib.layers.one_hot_encoding(
+                tf.constant([i]), num_classes=w),
+            [h, 1]) for i in range(w)
     ]
     w_loc = tf.concat([tf.expand_dims(t, 2) for t in w_loc], 2)
     loc = tf.concat([h_loc, w_loc], 2)
@@ -197,11 +215,12 @@ class ModelTest(tf.test.TestCase):
     conv_w_coords_tf = model.encode_coordinates_fn(self.fake_conv_tower)
 
     with self.test_session() as sess:
-      conv_w_coords = sess.run(conv_w_coords_tf)
+      conv_w_coords = sess.run(
+          conv_w_coords_tf, feed_dict={self.input_images: self.fake_images})
 
     batch_size, height, width, feature_size = self.conv_tower_shape
-    self.assertEqual(conv_w_coords.shape, (batch_size, height, width,
-                                           feature_size + height + width))
+    self.assertEqual(conv_w_coords.shape,
+                     (batch_size, height, width, feature_size + height + width))
 
   def test_disabled_coordinate_encoding_returns_features_unchanged(self):
     model = self.create_model()
@@ -209,7 +228,8 @@ class ModelTest(tf.test.TestCase):
     conv_w_coords_tf = model.encode_coordinates_fn(self.fake_conv_tower)
 
     with self.test_session() as sess:
-      conv_w_coords = sess.run(conv_w_coords_tf)
+      conv_w_coords = sess.run(
+          conv_w_coords_tf, feed_dict={self.input_images: self.fake_images})
 
     self.assertAllEqual(conv_w_coords, self.fake_conv_tower_np)
 
@@ -221,7 +241,8 @@ class ModelTest(tf.test.TestCase):
     conv_w_coords_tf = model.encode_coordinates_fn(fake_conv_tower)
 
     with self.test_session() as sess:
-      conv_w_coords = sess.run(conv_w_coords_tf)
+      conv_w_coords = sess.run(
+          conv_w_coords_tf, feed_dict={self.input_images: self.fake_images})
 
     # Original features
     self.assertAllEqual(conv_w_coords[0, :, :, :4],
@@ -252,8 +273,8 @@ class ModelTest(tf.test.TestCase):
       endpoints_tf = ocr_model.create_base(
           images=self.fake_images, labels_one_hot=None)
 
-      sess.run(tf.global_variables_initializer())
-      tf.tables_initializer().run()
+      sess.run(tf.compat.v1.global_variables_initializer())
+      tf.compat.v1.tables_initializer().run()
       endpoints = sess.run(endpoints_tf)
 
       self.assertEqual(endpoints.predicted_text.shape, (self.batch_size,))
@@ -261,14 +282,15 @@ class ModelTest(tf.test.TestCase):
 
 
 class CharsetMapperTest(tf.test.TestCase):
+
   def test_text_corresponds_to_ids(self):
     charset = create_fake_charset(36)
-    ids = tf.constant(
-        [[17, 14, 21, 21, 24], [32, 24, 27, 21, 13]], dtype=tf.int64)
+    ids = tf.constant([[17, 14, 21, 21, 24], [32, 24, 27, 21, 13]],
+                      dtype=tf.int64)
     charset_mapper = model.CharsetMapper(charset)
 
     with self.test_session() as sess:
-      tf.tables_initializer().run()
+      tf.compat.v1.tables_initializer().run()
       text = sess.run(charset_mapper.get_text(ids))
 
     self.assertAllEqual(text, [b'hello', b'world'])

research/attention_ocr/python/sequence_layers.py

@@ -111,12 +111,12 @@ class SequenceLayerBase(object):
     self._mparams = method_params
     self._net = net
     self._labels_one_hot = labels_one_hot
-    self._batch_size = net.get_shape().dims[0].value
+    self._batch_size = tf.shape(input=net)[0]
 
     # Initialize parameters for char logits which will be computed on the fly
     # inside an LSTM decoder.
     self._char_logits = {}
-    regularizer = slim.l2_regularizer(self._mparams.weight_decay)
+    regularizer = tf.keras.regularizers.l2(0.5 * (self._mparams.weight_decay))
     self._softmax_w = slim.model_variable(
         'softmax_w',
         [self._mparams.num_lstm_units, self._params.num_char_classes],
@@ -124,7 +124,7 @@ class SequenceLayerBase(object):
         regularizer=regularizer)
     self._softmax_b = slim.model_variable(
         'softmax_b', [self._params.num_char_classes],
-        initializer=tf.zeros_initializer(),
+        initializer=tf.compat.v1.zeros_initializer(),
         regularizer=regularizer)
 
   @abc.abstractmethod
@@ -203,8 +203,8 @@ class SequenceLayerBase(object):
       A tensor with shape [batch_size, num_char_classes]
     """
     if char_index not in self._char_logits:
-      self._char_logits[char_index] = tf.nn.xw_plus_b(inputs, self._softmax_w,
-                                                      self._softmax_b)
+      self._char_logits[char_index] = tf.compat.v1.nn.xw_plus_b(inputs, self._softmax_w,
+                                                                self._softmax_b)
     return self._char_logits[char_index]
 
   def char_one_hot(self, logit):
@@ -216,7 +216,7 @@ class SequenceLayerBase(object):
     Returns:
       A tensor with shape [batch_size, num_char_classes]
     """
-    prediction = tf.argmax(logit, axis=1)
+    prediction = tf.argmax(input=logit, axis=1)
     return slim.one_hot_encoding(prediction, self._params.num_char_classes)
 
   def get_input(self, prev, i):
@@ -244,10 +244,10 @@ class SequenceLayerBase(object):
     Returns:
       A tensor with shape [batch_size, seq_length, num_char_classes].
     """
-    with tf.variable_scope('LSTM'):
+    with tf.compat.v1.variable_scope('LSTM'):
       first_label = self.get_input(prev=None, i=0)
       decoder_inputs = [first_label] + [None] * (self._params.seq_length - 1)
-      lstm_cell = tf.contrib.rnn.LSTMCell(
+      lstm_cell = tf.compat.v1.nn.rnn_cell.LSTMCell(
           self._mparams.num_lstm_units,
           use_peepholes=False,
           cell_clip=self._mparams.lstm_state_clip_value,
@@ -259,9 +259,9 @@ class SequenceLayerBase(object):
           loop_function=self.get_input,
           cell=lstm_cell)
 
-    with tf.variable_scope('logits'):
+    with tf.compat.v1.variable_scope('logits'):
       logits_list = [
-          tf.expand_dims(self.char_logit(logit, i), dim=1)
+          tf.expand_dims(self.char_logit(logit, i), axis=1)
           for i, logit in enumerate(lstm_outputs)
       ]
 
@@ -275,7 +275,7 @@ class NetSlice(SequenceLayerBase):
   def __init__(self, *args, **kwargs):
     super(NetSlice, self).__init__(*args, **kwargs)
     self._zero_label = tf.zeros(
-        [self._batch_size, self._params.num_char_classes])
+        tf.stack([self._batch_size, self._params.num_char_classes]))
 
   def get_image_feature(self, char_index):
     """Returns a subset of image features for a character.
@@ -352,7 +352,7 @@ class Attention(SequenceLayerBase):
   def __init__(self, *args, **kwargs):
     super(Attention, self).__init__(*args, **kwargs)
     self._zero_label = tf.zeros(
-        [self._batch_size, self._params.num_char_classes])
+        tf.stack([self._batch_size, self._params.num_char_classes]))
 
   def get_eval_input(self, prev, i):
     """See SequenceLayerBase.get_eval_input for details."""

research/attention_ocr/python/sequence_layers_test.py

@@ -29,13 +29,13 @@ import sequence_layers
 
 def fake_net(batch_size, num_features, feature_size):
   return tf.convert_to_tensor(
-      np.random.uniform(size=(batch_size, num_features, feature_size)),
+      value=np.random.uniform(size=(batch_size, num_features, feature_size)),
       dtype=tf.float32)
 
 
 def fake_labels(batch_size, seq_length, num_char_classes):
   labels_np = tf.convert_to_tensor(
-      np.random.randint(
+      value=np.random.randint(
           low=0, high=num_char_classes, size=(batch_size, seq_length)))
   return slim.one_hot_encoding(labels_np, num_classes=num_char_classes)
 

research/attention_ocr/python/train.py

@@ -96,16 +96,16 @@ def get_training_hparams():
 def create_optimizer(hparams):
   """Creates optimized based on the specified flags."""
   if hparams.optimizer == 'momentum':
-    optimizer = tf.train.MomentumOptimizer(
+    optimizer = tf.compat.v1.train.MomentumOptimizer(
         hparams.learning_rate, momentum=hparams.momentum)
   elif hparams.optimizer == 'adam':
-    optimizer = tf.train.AdamOptimizer(hparams.learning_rate)
+    optimizer = tf.compat.v1.train.AdamOptimizer(hparams.learning_rate)
   elif hparams.optimizer == 'adadelta':
-    optimizer = tf.train.AdadeltaOptimizer(hparams.learning_rate)
+    optimizer = tf.compat.v1.train.AdadeltaOptimizer(hparams.learning_rate)
   elif hparams.optimizer == 'adagrad':
-    optimizer = tf.train.AdagradOptimizer(hparams.learning_rate)
+    optimizer = tf.compat.v1.train.AdagradOptimizer(hparams.learning_rate)
   elif hparams.optimizer == 'rmsprop':
-    optimizer = tf.train.RMSPropOptimizer(
+    optimizer = tf.compat.v1.train.RMSPropOptimizer(
         hparams.learning_rate, momentum=hparams.momentum)
   return optimizer
 
@@ -154,14 +154,14 @@ def train(loss, init_fn, hparams):
 
 
 def prepare_training_dir():
-  if not tf.gfile.Exists(FLAGS.train_log_dir):
+  if not tf.io.gfile.exists(FLAGS.train_log_dir):
     logging.info('Create a new training directory %s', FLAGS.train_log_dir)
-    tf.gfile.MakeDirs(FLAGS.train_log_dir)
+    tf.io.gfile.makedirs(FLAGS.train_log_dir)
   else:
     if FLAGS.reset_train_dir:
       logging.info('Reset the training directory %s', FLAGS.train_log_dir)
-      tf.gfile.DeleteRecursively(FLAGS.train_log_dir)
-      tf.gfile.MakeDirs(FLAGS.train_log_dir)
+      tf.io.gfile.rmtree(FLAGS.train_log_dir)
+      tf.io.gfile.makedirs(FLAGS.train_log_dir)
     else:
       logging.info('Use already existing training directory %s',
                    FLAGS.train_log_dir)
@@ -169,7 +169,7 @@ def prepare_training_dir():
 
 def calculate_graph_metrics():
   param_stats = model_analyzer.print_model_analysis(
-      tf.get_default_graph(),
+      tf.compat.v1.get_default_graph(),
       tfprof_options=model_analyzer.TRAINABLE_VARS_PARAMS_STAT_OPTIONS)
   return param_stats.total_parameters
 
@@ -186,7 +186,7 @@ def main(_):
   # If ps_tasks is zero, the local device is used. When using multiple
   # (non-local) replicas, the ReplicaDeviceSetter distributes the variables
   # across the different devices.
-  device_setter = tf.train.replica_device_setter(
+  device_setter = tf.compat.v1.train.replica_device_setter(
       FLAGS.ps_tasks, merge_devices=True)
   with tf.device(device_setter):
     data = data_provider.get_data(

research/attention_ocr/python/utils.py

@@ -37,16 +37,16 @@ def logits_to_log_prob(logits):
     probabilities.
   """
 
-  with tf.variable_scope('log_probabilities'):
+  with tf.compat.v1.variable_scope('log_probabilities'):
     reduction_indices = len(logits.shape.as_list()) - 1
     max_logits = tf.reduce_max(
-        logits, reduction_indices=reduction_indices, keep_dims=True)
+        input_tensor=logits, axis=reduction_indices, keepdims=True)
     safe_logits = tf.subtract(logits, max_logits)
     sum_exp = tf.reduce_sum(
-        tf.exp(safe_logits),
-        reduction_indices=reduction_indices,
-        keep_dims=True)
-    log_probs = tf.subtract(safe_logits, tf.log(sum_exp))
+        input_tensor=tf.exp(safe_logits),
+        axis=reduction_indices,
+        keepdims=True)
+    log_probs = tf.subtract(safe_logits, tf.math.log(sum_exp))
   return log_probs
 
 
@@ -78,3 +78,20 @@ def variables_to_restore(scope=None, strip_scope=False):
     return variable_map
   else:
     return {v.op.name: v for v in slim.get_variables_to_restore()}
+
+
+def ConvertAllInputsToTensors(func):
+  """A decorator to convert all function's inputs into tensors.
+
+  Args:
+    func: a function to decorate.
+
+  Returns:
+    A decorated function.
+  """
+
+  def FuncWrapper(*args):
+    tensors = [tf.convert_to_tensor(value=a) for a in args]
+    return func(*tensors)
+
+  return FuncWrapper

research/deeplab/datasets/build_cityscapes_data.py

@@ -113,17 +113,23 @@ def _get_files(data, dataset_split):
 
   Args:
     data: String, desired data ('image' or 'label').
-    dataset_split: String, dataset split ('train', 'val', 'test')
+    dataset_split: String, dataset split ('train_fine', 'val_fine', 'test_fine')
 
   Returns:
     A list of sorted file names or None when getting label for
       test set.
   """
-  if data == 'label' and dataset_split == 'test':
-    return None
+  if dataset_split == 'train_fine':
+    split_dir = 'train'
+  elif dataset_split == 'val_fine':
+    split_dir = 'val'
+  elif dataset_split == 'test_fine':
+    split_dir = 'test'
+  else:
+    raise RuntimeError("Split {} is not supported".format(dataset_split))
   pattern = '*%s.%s' % (_POSTFIX_MAP[data], _DATA_FORMAT_MAP[data])
   search_files = os.path.join(
-      FLAGS.cityscapes_root, _FOLDERS_MAP[data], dataset_split, '*', pattern)
+      FLAGS.cityscapes_root, _FOLDERS_MAP[data], split_dir, '*', pattern)
   filenames = glob.glob(search_files)
   return sorted(filenames)
 
@@ -132,7 +138,7 @@ def _convert_dataset(dataset_split):
   """Converts the specified dataset split to TFRecord format.
 
   Args:
-    dataset_split: The dataset split (e.g., train, val).
+    dataset_split: The dataset split (e.g., train_fine, val_fine).
 
   Raises:
     RuntimeError: If loaded image and label have different shape, or if the
@@ -142,8 +148,12 @@ def _convert_dataset(dataset_split):
   label_files = _get_files('label', dataset_split)
 
   num_images = len(image_files)
+  num_labels = len(label_files)
   num_per_shard = int(math.ceil(num_images / _NUM_SHARDS))
 
+  if num_images != num_labels:
+    raise RuntimeError("The number of images and labels doesn't match: {} {}".format(num_images, num_labels))
+
   image_reader = build_data.ImageReader('png', channels=3)
   label_reader = build_data.ImageReader('png', channels=1)
 
@@ -179,8 +189,8 @@ def _convert_dataset(dataset_split):
 
 
 def main(unused_argv):
-  # Only support converting 'train' and 'val' sets for now.
-  for dataset_split in ['train', 'val']:
+  # Only support converting 'train_fine', 'val_fine' and 'test_fine' sets for now.
+  for dataset_split in ['train_fine', 'val_fine', 'test_fine']:
     _convert_dataset(dataset_split)
 
 

research/deeplab/datasets/convert_cityscapes.sh

@@ -42,6 +42,8 @@ WORK_DIR="."
 # Root path for Cityscapes dataset.
 CITYSCAPES_ROOT="${WORK_DIR}/cityscapes"
 
+export PYTHONPATH="${CITYSCAPES_ROOT}:${PYTHONPATH}"
+
 # Create training labels.
 python "${CITYSCAPES_ROOT}/cityscapesscripts/preparation/createTrainIdLabelImgs.py"
 

research/deeplab/deprecated/segmentation_dataset.py

@@ -81,8 +81,8 @@ DatasetDescriptor = collections.namedtuple(
 
 _CITYSCAPES_INFORMATION = DatasetDescriptor(
     splits_to_sizes={
-        'train': 2975,
-        'val': 500,
+        'train_fine': 2975,
+        'val_fine': 500,
     },
     num_classes=19,
     ignore_label=255,

research/deeplab/g3doc/cityscapes.md

@@ -43,7 +43,7 @@ A local training job using `xception_65` can be run with the following command:
 python deeplab/train.py \
     --logtostderr \
     --training_number_of_steps=90000 \
-    --train_split="train" \
+    --train_split="train_fine" \
     --model_variant="xception_65" \
     --atrous_rates=6 \
     --atrous_rates=12 \
@@ -95,7 +95,7 @@ command:
 # From tensorflow/models/research/
 python deeplab/eval.py \
     --logtostderr \
-    --eval_split="val" \
+    --eval_split="val_fine" \
     --model_variant="xception_65" \
     --atrous_rates=6 \
     --atrous_rates=12 \
@@ -121,7 +121,7 @@ command:
 # From tensorflow/models/research/
 python deeplab/vis.py \
     --logtostderr \
-    --vis_split="val" \
+    --vis_split="val_fine" \
     --model_variant="xception_65" \
     --atrous_rates=6 \
     --atrous_rates=12 \

research/deeplab/g3doc/installation.md

@@ -68,6 +68,6 @@ Quick running the whole code on the PASCAL VOC 2012 dataset:
 
 ```bash
 # From tensorflow/models/research/deeplab
-sh local_test.sh
+bash local_test.sh
 ```
 

research/deeplab/local_test.sh

@@ -19,7 +19,7 @@
 #
 # Usage:
 #   # From the tensorflow/models/research/deeplab directory.
-#   sh ./local_test.sh
+#   bash ./local_test.sh
 #
 #
 
@@ -42,7 +42,7 @@ python "${WORK_DIR}"/model_test.py
 # Go to datasets folder and download PASCAL VOC 2012 segmentation dataset.
 DATASET_DIR="datasets"
 cd "${WORK_DIR}/${DATASET_DIR}"
-sh download_and_convert_voc2012.sh
+bash download_and_convert_voc2012.sh
 
 # Go back to original directory.
 cd "${CURRENT_DIR}"

research/delf/INSTALL_INSTRUCTIONS.md

 ## DELF installation
 
+### Installation script
+
+We now have a script to do the entire installation in one shot. Navigate to the
+directory `models/research/delf/delf/python/training`, then run:
+
+```bash
+# From models/research/delf/delf/python/training
+bash install_delf.sh
+```
+
+If this works, you are done! If not, see below for detailed instructions for
+installing this codebase and its dependencies.
+
+*Please note that this installation script only works on 64 bits Linux
+architectures due to the `protoc` binary that is automatically downloaded. If
+you wish to install the DELF library on other architectures please update the
+[`install_delf.sh`](delf/python/training/install_delf.sh) script by referencing
+the desired `protoc`
+[binary release](https://github.com/protocolbuffers/protobuf/releases).*
+
+In more detail: the `install_delf.sh` script installs both the DELF library and
+its dependencies in the following sequence:
+
+*   Install TensorFlow 2.2 and TensorFlow 2.2 for GPU.
+*   Install the [TF-Slim](https://github.com/google-research/tf-slim) library
+    from source.
+*   Download [protoc](https://github.com/protocolbuffers/protobuf) and compile
+    the DELF Protocol Buffers.
+*   Install the matplotlib, numpy, scikit-image, scipy and python3-tk Python
+    libraries.
+*   Install the
+    [TensorFlow Object Detection API](https://github.com/tensorflow/models/tree/master/research/object_detection)
+    from the cloned TensorFlow Model Garden repository.
+*   Install the DELF package.
+
 ### Tensorflow
 
-[![TensorFlow 2.1](https://img.shields.io/badge/tensorflow-2.1-brightgreen)](https://github.com/tensorflow/tensorflow/releases/tag/v2.1.0)
+[![TensorFlow 2.2](https://img.shields.io/badge/tensorflow-2.2-brightgreen)](https://github.com/tensorflow/tensorflow/releases/tag/v2.2.0)
 [![Python 3.6](https://img.shields.io/badge/python-3.6-blue.svg)](https://www.python.org/downloads/release/python-360/)
 
 For detailed steps to install Tensorflow, follow the
@@ -11,9 +46,9 @@ typical user can install Tensorflow using one of the following commands:
 
 ```bash
 # For CPU:
-pip3 install 'tensorflow'
+pip3 install 'tensorflow>=2.2.0'
 # For GPU:
-pip3 install 'tensorflow-gpu'
+pip3 install 'tensorflow-gpu>=2.2.0'
 ```
 
 ### TF-Slim

research/delf/README.md

 # Deep Local and Global Image Features
 
-[![TensorFlow 2.1](https://img.shields.io/badge/tensorflow-2.1-brightgreen)](https://github.com/tensorflow/tensorflow/releases/tag/v2.1.0)
+[![TensorFlow 2.2](https://img.shields.io/badge/tensorflow-2.2-brightgreen)](https://github.com/tensorflow/tensorflow/releases/tag/v2.2.0)
 [![Python 3.6](https://img.shields.io/badge/python-3.6-blue.svg)](https://www.python.org/downloads/release/python-360/)
 
 This project presents code for deep local and global image feature methods,
@@ -41,7 +41,7 @@ DELG:
 ```
 "Unifying Deep Local and Global Features for Image Search",
 B. Cao*, A. Araujo* and J. Sim,
-arxiv:2001.05027
+Proc. ECCV'20
 ```
 
 GLDv2:
@@ -55,11 +55,11 @@ Proc. CVPR'20
 
 ## News
 
+-   [Jul'20] Check out our ECCV'20 paper:
+    ["Unifying Deep Local and Global Features for Image Search"](https://arxiv.org/abs/2001.05027)
 -   [Apr'20] Check out our CVPR'20 paper: ["Google Landmarks Dataset v2 - A
     Large-Scale Benchmark for Instance-Level Recognition and
     Retrieval"](https://arxiv.org/abs/2004.01804)
--   [Jan'20] Check out our new paper:
-    ["Unifying Deep Local and Global Features for Image Search"](https://arxiv.org/abs/2001.05027)
 -   [Jun'19] DELF achieved 2nd place in
     [CVPR Visual Localization challenge (Local Features track)](https://sites.google.com/corp/view/ltvl2019).
     See our slides
@@ -182,104 +182,55 @@ directories therein, `protos` and `python`.
 
 ### `delf/protos`
 
-This directory contains protobufs:
-
--   `aggregation_config.proto`: protobuf for configuring local feature
-    aggregation.
--   `box.proto`: protobuf for serializing detected boxes.
--   `datum.proto`: general-purpose protobuf for serializing float tensors.
--   `delf_config.proto`: protobuf for configuring DELF/DELG extraction.
--   `feature.proto`: protobuf for serializing DELF features.
+This directory contains protobufs for local feature aggregation
+(`aggregation_config.proto`), serializing detected boxes (`box.proto`),
+serializing float tensors (`datum.proto`), configuring DELF/DELG extraction
+(`delf_config.proto`), serializing local features (`feature.proto`).
 
 ### `delf/python`
 
-This directory contains files for several different purposes:
-
--   `box_io.py`, `datum_io.py`, `feature_io.py` are helper files for reading and
-    writing tensors and features.
--   `delf_v1.py` contains code to create DELF models.
--   `feature_aggregation_extractor.py` contains a module to perform local
-    feature aggregation.
--   `feature_aggregation_similarity.py` contains a module to perform similarity
-    computation for aggregated local features.
--   `feature_extractor.py` contains the code to extract features using DELF.
-    This is particularly useful for extracting features over multiple scales,
-    with keypoint selection based on attention scores, and PCA/whitening
-    post-processing.
-
-The subdirectory `delf/python/examples` contains sample scripts to run DELF
-feature extraction/matching, and object detection:
-
--   `delf_config_example.pbtxt` shows an example instantiation of the DelfConfig
-    proto, used for DELF feature extraction.
--   `detector.py` is a module to construct an object detector function.
--   `extract_boxes.py` enables object detection from a list of images.
--   `extract_features.py` enables DELF extraction from a list of images.
--   `extractor.py` is a module to construct a DELF/DELG local feature extraction
-    function.
--   `match_images.py` supports image matching using DELF features extracted
-    using `extract_features.py`.
+This directory contains files for several different purposes, such as:
+reading/writing tensors/features (`box_io.py`, `datum_io.py`, `feature_io.py`),
+local feature aggregation extraction and similarity computation
+(`feature_aggregation_extractor.py`, `feature_aggregation_similarity.py`) and
+helper functions for image/feature loading/processing (`utils.py`,
+`feature_extractor.py`).
 
-The subdirectory `delf/python/delg` contains sample scripts/configs related to
-the DELG paper:
+The subdirectory `delf/python/examples` contains sample scripts to run DELF/DELG
+feature extraction/matching (`extractor.py`, `extract_features.py`,
+`match_images.py`) and object detection (`detector.py`, `extract_boxes.py`).
+`delf_config_example.pbtxt` shows an example instantiation of the DelfConfig
+proto, used for DELF feature extraction.
 
--   `delg_gld_config.pbtxt` gives the DelfConfig used in DELG paper.
--   `extract_features.py` for local+global feature extraction on Revisited
-    datasets.
--   `perform_retrieval.py` for performing retrieval/evaluating methods on
-    Revisited datasets.
+The subdirectory `delf/python/delg` contains sample scripts/configs related to
+the DELG paper: `extract_features.py` for local+global feature extraction (with
+and example `delg_gld_config.pbtxt`) and `perform_retrieval.py` for performing
+retrieval/scoring.
 
 The subdirectory `delf/python/detect_to_retrieve` contains sample
-scripts/configs related to the Detect-to-Retrieve paper:
-
--   `aggregation_extraction.py` is a library to extract/save feature
-    aggregation.
--   `boxes_and_features_extraction.py` is a library to extract/save boxes and
-    DELF features.
--   `cluster_delf_features.py` for local feature clustering.
--   `dataset.py` for parsing/evaluating results on Revisited Oxford/Paris
-    datasets.
--   `delf_gld_config.pbtxt` gives the DelfConfig used in Detect-to-Retrieve
-    paper.
--   `extract_aggregation.py` for aggregated local feature extraction.
--   `extract_index_boxes_and_features.py` for index image local feature
-    extraction / bounding box detection on Revisited datasets.
--   `extract_query_features.py` for query image local feature extraction on
-    Revisited datasets.
--   `image_reranking.py` is a module to re-rank images with geometric
-    verification.
--   `perform_retrieval.py` for performing retrieval/evaluating methods using
-    aggregated local features on Revisited datasets.
--   `index_aggregation_config.pbtxt`, `query_aggregation_config.pbtxt` give
-    AggregationConfig's for Detect-to-Retrieve experiments.
+scripts/configs related to the Detect-to-Retrieve paper, for feature/box
+extraction/aggregation/clustering (`aggregation_extraction.py`,
+`boxes_and_features_extraction.py`, `cluster_delf_features.py`,
+`extract_aggregation.py`, `extract_index_boxes_and_features.py`,
+`extract_query_features.py`), image retrieval/reranking (`perform_retrieval.py`,
+`image_reranking.py`), along with configs used for feature
+extraction/aggregation (`delf_gld_config.pbtxt`,
+`index_aggregation_config.pbtxt`, `query_aggregation_config.pbtxt`) and
+Revisited Oxford/Paris dataset parsing/evaluation (`dataset.py`).
 
 The subdirectory `delf/python/google_landmarks_dataset` contains sample
-scripts/modules for computing GLD metrics / reproducing results from the GLDv2
-paper:
-
--   `compute_recognition_metrics.py` performs recognition metric computation
-    given input predictions and solution files.
--   `compute_retrieval_metrics.py` performs retrieval metric computation given
-    input predictions and solution files.
--   `dataset_file_io.py` is a module for dataset-related file IO.
--   `metrics.py` is a module for GLD metric computation.
--   `rn101_af_gldv2clean_config.pbtxt` gives the DelfConfig used in the
-    ResNet101-ArcFace (trained on GLDv2-train-clean) baseline used in the GLDv2
-    paper.
+scripts/modules for computing GLD metrics (`metrics.py`,
+`compute_recognition_metrics.py`, `compute_retrieval_metrics.py`), GLD file IO
+(`dataset_file_io.py`) / reproducing results from the GLDv2 paper
+(`rn101_af_gldv2clean_config.pbtxt` and the instructions therein).
 
 The subdirectory `delf/python/training` contains sample scripts/modules for
-performing DELF training:
-
--   `datasets/googlelandmarks.py` is the dataset module used for training.
--   `model/delf_model.py` is the model module used for training.
--   `model/export_model.py` is a script for exporting trained models in the
-    format used by the inference code.
--   `model/export_model_utils.py` is a module with utilities for model
-    exporting.
--   `model/resnet50.py` is a module with a backbone RN50 implementation.
--   `build_image_dataset.py` converts downloaded dataset into TFRecords format
-    for training.
--   `train.py` is the main training script.
+performing model training (`train.py`) based on a ResNet50 DELF model
+(`model/resnet50.py`, `model/delf_model.py`), also presenting relevant model
+exporting scripts and associated utils (`model/export_model.py`,
+`model/export_global_model.py`, `model/export_model_utils.py`) and dataset
+downloading/preprocessing (`download_dataset.sh`, `build_image_dataset.py`,
+`datasets/googlelandmarks.py`).
 
 Besides these, other files in the different subdirectories contain tests for the
 various modules.
@@ -290,6 +241,16 @@ André Araujo (@andrefaraujo)
 
 ## Release history
 
+### Jul, 2020
+
+-   Full TF2 support. Only one minor `compat.v1` usage left. Updated
+    instructions to require TF2.2
+-   Refactored / much improved training code, with very detailed, step-by-step
+    instructions
+
+**Thanks to contributors**: Dan Anghel, Barbara Fusinska and André
+Araujo.
+
 ### May, 2020
 
 -   Codebase is now Python3-first

research/delf/delf/python/training/README.md

@@ -24,34 +24,9 @@ cd models/research/delf/delf/python/training
 
 ## Install the DELF Library
 
-The DELF Python library can be installed by running the
-[`install_delf.sh`](./install_delf.sh) script using the command:
-
-```
-bash install_delf.sh
-```
-
-The script installs both the DELF library and its dependencies in the following
-sequence:
-
-*   Install TensorFlow 2.2 and TensorFlow 2.2 for GPU.
-*   Install the [TF-Slim](https://github.com/google-research/tf-slim) library
-    from source.
-*   Download [protoc](https://github.com/protocolbuffers/protobuf) and compile
-    the DELF Protocol Buffers.
-*   Install the matplotlib, numpy, scikit-image, scipy and python3-tk Python
-    libraries.
-*   Install the
-    [TensorFlow Object Detection API](https://github.com/tensorflow/models/tree/master/research/object_detection)
-    from the cloned TensorFlow Model Garden repository.
-*   Install the DELF package.
-
-*Please note that the current installation only works on 64 bits Linux
-architectures due to the `protoc` binary downloaded by the installation script.
-If you wish to install the DELF library on other architectures please update the
-[`install_delf.sh`](./install_delf.sh) script by referencing the desired
-`protoc`
-[binary release](https://github.com/protocolbuffers/protobuf/releases).*
+To be able to use this code, please follow
+[these instructions](../../../INSTALL_INSTRUCTIONS.md) to properly install the
+DELF library.
 
 ## Download the GLDv2 Training Data