tf.ensure_shape

Updates the shape of a tensor and checks at runtime that the shape holds.

When executed, this operation asserts that the input tensor x's shape is compatible with the shape argument. See tf.TensorShape.is_compatible_with for details.

x = tf.constant([[1, 2, 3],
                 [4, 5, 6]])
x = tf.ensure_shape(x, [2, 3])

Use None for unknown dimensions:

x = tf.ensure_shape(x, [None, 3])
x = tf.ensure_shape(x, [2, None])

If the tensor's shape is not compatible with the shape argument, an error is raised:

x = tf.ensure_shape(x, [5])
Traceback (most recent call last):

tf.errors.InvalidArgumentError: Shape of tensor dummy_input [3] is not
  compatible with expected shape [5]. [Op:EnsureShape]

During graph construction (typically tracing a tf.function), tf.ensure_shape updates the static-shape of the result tensor by merging the two shapes. See tf.TensorShape.merge_with for details.

This is most useful when you know a shape that can't be determined statically by TensorFlow.

The following trivial tf.function prints the input tensor's static-shape before and after ensure_shape is applied.

@tf.function
def f(tensor):
  print("Static-shape before:", tensor.shape)
  tensor = tf.ensure_shape(tensor, [None, 3])
  print("Static-shape after:", tensor.shape)
  return tensor

This lets you see the effect of tf.ensure_shape when the function is traced:

>>> cf = f.get_concrete_function(tf.TensorSpec([None, None]))
Static-shape before: (None, None)
Static-shape after: (None, 3)
cf(tf.zeros([3, 3])) # Passes
cf(tf.constant([1, 2, 3])) # fails
Traceback (most recent call last):

InvalidArgumentError:  Shape of tensor x [3] is not compatible with expected shape [3,3].

The above example raises tf.errors.InvalidArgumentError, because x's shape, (3,), is not compatible with the shape argument, (None, 3)

Inside a tf.function or v1.Graph context it checks both the buildtime and runtime shapes. This is stricter than tf.Tensor.set_shape which only checks the buildtime shape.

For example, of loading images of a known size:

@tf.function
def decode_image(png):
  image = tf.image.decode_png(png, channels=3)
  # the `print` executes during tracing.
  print("Initial shape: ", image.shape)
  image = tf.ensure_shape(image,[28, 28, 3])
  print("Final shape: ", image.shape)
  return image

When tracing a function, no ops are being executed, shapes may be unknown. See the Concrete Functions Guide for details.

concrete_decode = decode_image.get_concrete_function(
    tf.TensorSpec([], dtype=tf.string))
Initial shape:  (None, None, 3)
Final shape:  (28, 28, 3)
image = tf.random.uniform(maxval=255, shape=[28, 28, 3], dtype=tf.int32)
image = tf.cast(image,tf.uint8)
png = tf.image.encode_png(image)
image2 = concrete_decode(png)
print(image2.shape)
(28, 28, 3)
image = tf.concat([image,image], axis=0)
print(image.shape)
(56, 28, 3)
png = tf.image.encode_png(image)
image2 = concrete_decode(png)
Traceback (most recent call last):

tf.errors.InvalidArgumentError:  Shape of tensor DecodePng [56,28,3] is not
  compatible with expected shape [28,28,3].
@tf.function
def bad_decode_image(png):
  image = tf.image.decode_png(png, channels=3)
  # the `print` executes during tracing.
  print("Initial shape: ", image.shape)
  # BAD: forgot to use the returned tensor.