def testBuildSignatureDef(self):
# Force the test to run in graph mode.
# This tests a deprecated v1 API that uses functionality that does not work
# with eager tensors (namely build_tensor_info).
with ops.Graph().as_default():
x = array_ops.placeholder(dtypes.float32, 1, name="x")
x_tensor_info = utils.build_tensor_info(x)
inputs = {}
inputs["foo-input"] = x_tensor_info
y = array_ops.placeholder(dtypes.float32, name="y")
y_tensor_info = utils.build_tensor_info(y)
outputs = {}
outputs["foo-output"] = y_tensor_info
signature_def = signature_def_utils_impl.build_signature_def(
inputs, outputs, "foo-method-name")
self.assertEqual("foo-method-name", signature_def.method_name)
# Check inputs in signature def.
self.assertEqual(1, len(signature_def.inputs))
x_tensor_info_actual = signature_def.inputs["foo-input"]
self.assertEqual("x:0", x_tensor_info_actual.name)
self.assertEqual(types_pb2.DT_FLOAT, x_tensor_info_actual.dtype)
self.assertEqual(1, len(x_tensor_info_actual.tensor_shape.dim))
self.assertEqual(1, x_tensor_info_actual.tensor_shape.dim[0].size)
# Check outputs in signature def.
self.assertEqual(1, len(signature_def.outputs))
y_tensor_info_actual = signature_def.outputs["foo-output"]
self.assertEqual("y:0", y_tensor_info_actual.name)
self.assertEqual(types_pb2.DT_FLOAT, y_tensor_info_actual.dtype)
self.assertEqual(0, len(y_tensor_info_actual.tensor_shape.dim))
def testBuildSignatureDef(self):
x = array_ops.placeholder(dtypes.float32, 1, name="x")
x_tensor_info = utils.build_tensor_info(x)
inputs = {}
inputs["foo-input"] = x_tensor_info
y = array_ops.placeholder(dtypes.float32, name="y")
y_tensor_info = utils.build_tensor_info(y)
outputs = {}
outputs["foo-output"] = y_tensor_info
signature_def = signature_def_utils_impl.build_signature_def(
inputs, outputs, "foo-method-name")
self.assertEqual("foo-method-name", signature_def.method_name)
# Check inputs in signature def.
self.assertEqual(1, len(signature_def.inputs))
x_tensor_info_actual = signature_def.inputs["foo-input"]
self.assertEqual("x:0", x_tensor_info_actual.name)
self.assertEqual(types_pb2.DT_FLOAT, x_tensor_info_actual.dtype)
self.assertEqual(1, len(x_tensor_info_actual.tensor_shape.dim))
self.assertEqual(1, x_tensor_info_actual.tensor_shape.dim[0].size)
# Check outputs in signature def.
self.assertEqual(1, len(signature_def.outputs))
y_tensor_info_actual = signature_def.outputs["foo-output"]
self.assertEqual("y:0", y_tensor_info_actual.name)
self.assertEqual(types_pb2.DT_FLOAT, y_tensor_info_actual.dtype)
self.assertEqual(0, len(y_tensor_info_actual.tensor_shape.dim))
def testBuildSignatureDef(self):
x = array_ops.placeholder(dtypes.float32, 1, name="x")
x_tensor_info = utils.build_tensor_info(x)
inputs = {}
inputs["foo-input"] = x_tensor_info
y = array_ops.placeholder(dtypes.float32, name="y")
y_tensor_info = utils.build_tensor_info(y)
outputs = {}
outputs["foo-output"] = y_tensor_info
signature_def = signature_def_utils_impl.build_signature_def(
inputs, outputs, "foo-method-name")
self.assertEqual("foo-method-name", signature_def.method_name)
# Check inputs in signature def.
self.assertEqual(1, len(signature_def.inputs))
x_tensor_info_actual = signature_def.inputs["foo-input"]
self.assertEqual("x:0", x_tensor_info_actual.name)
self.assertEqual(types_pb2.DT_FLOAT, x_tensor_info_actual.dtype)
self.assertEqual(1, len(x_tensor_info_actual.tensor_shape.dim))
self.assertEqual(1, x_tensor_info_actual.tensor_shape.dim[0].size)
# Check outputs in signature def.
self.assertEqual(1, len(signature_def.outputs))
y_tensor_info_actual = signature_def.outputs["foo-output"]
self.assertEqual("y:0", y_tensor_info_actual.name)
self.assertEqual(types_pb2.DT_FLOAT, y_tensor_info_actual.dtype)
self.assertEqual(0, len(y_tensor_info_actual.tensor_shape.dim))
def _make_signature(inputs, outputs, name=None):
input_info = {
input_name: utils.build_tensor_info(tensor)
for input_name, tensor in inputs.items()
}
output_info = {
output_name: utils.build_tensor_info(tensor)
for output_name, tensor in outputs.items()
}
return signature_def_utils_impl.build_signature_def(
input_info, output_info, name)
def _make_signature(inputs, outputs, name=None):
input_info = {
input_name: utils.build_tensor_info(tensor)
for input_name, tensor in inputs.items()
}
output_info = {
output_name: utils.build_tensor_info(tensor)
for output_name, tensor in outputs.items()
}
return signature_def_utils_impl.build_signature_def(input_info, output_info,
name)
def main(_):
print("Start the training...")
#import model
model = VGG16(weights = 'imagenet', include_top = True)
sgd = SGD(lr = 0.1, decay = 1e-6, momentum = 0.9, nesterov = True)
#compile the model
model.compile(optimizer = sgd, loss = 'categorical_crossentropy')
#view the properties of the model
model.summary()
#get input layer
input_layer = model.get_layer('input_1')
input_layer_in = input_layer.input
print("The input layer: ", input_layer_in)
#get predictions layer
predictions_layer = model.get_layer('predictions')
predictions_layer_out = predictions_layer.output
print("The predictions layer: ", predictions_layer_out)
#convert input layer and predictions layer from tensor to tensor info
input_layer_in_info = build_tensor_info(input_layer_in)
predictions_layer_out_info = build_tensor_info(predictions_layer_out)
#export model
K.set_learning_phase(0)
export_path_base = "./vgg16"
model_version = "1"
export_path = os.path.join(
tf.compat.as_bytes(export_path_base),
tf.compat.as_bytes(model_version))
print("Exporting trained model to: ", export_path)
prediction_signature = build_signature_def(
inputs = {'images': input_layer_in_info},
outputs = {'scores': predictions_layer_out_info},
method_name = signature_constants.PREDICT_METHOD_NAME)
builder = saved_model_builder.SavedModelBuilder(export_path)
with K.get_session() as sess:
builder.add_meta_graph_and_variables(
sess = sess,
tags = [tag_constants.SERVING],
signature_def_map = {'predict_images': prediction_signature})
builder.save()
#clear session
K.clear_session()
print("Done exporting!")
def _create_signature(model, class_names: List[str], top_k: Optional[int]):
serialized_tf_example = array_ops.placeholder(tf.string, name="tf_example")
feature_configs = {"x": tf.io.FixedLenFeature([], tf.string)}
tf_example = tf.io.parse_example(serialized_tf_example, feature_configs)
jpegs = tf_example["x"]
x = tf.map_fn(_preprocess_image, jpegs, dtype=tf.float32)
y = model(x)
top_k = min(top_k or len(class_names), len(class_names))
values, indices = tf.nn.top_k(y, top_k)
table_class_names = lookup_ops.index_to_string_table_from_tensor(
vocabulary_list=tf.constant(class_names),
default_value="UNK",
name=None)
classification_inputs = build_tensor_info(serialized_tf_example)
prediction_class_names = table_class_names.lookup(
tf.cast(indices, dtype=dtypes.int64))
classification_outputs_class_names = build_tensor_info(
prediction_class_names)
classification_outputs_scores = build_tensor_info(values)
classification_signature = build_signature_def(
inputs={sig_consts_v1.CLASSIFY_INPUTS: classification_inputs},
outputs={
sig_consts_v1.CLASSIFY_OUTPUT_CLASSES:
classification_outputs_class_names,
sig_consts_v1.CLASSIFY_OUTPUT_SCORES:
classification_outputs_scores,
},
method_name=sig_consts_v1.CLASSIFY_METHOD_NAME,
)
# Ensure valid signature
if not is_valid_signature(classification_signature):
raise ValueError("Invalid classification signature!")
return classification_signature
def _assertInvalidSignature(self, inputs, outputs, method_name):
signature_def = signature_def_utils_impl.build_signature_def(
inputs, outputs, method_name)
self.assertFalse(
signature_def_utils_impl.is_valid_signature(signature_def))
def main(_):
print("Start the training...")
#import model
model = VGG16(weights='imagenet',
include_top=True,
input_shape=(224, 224, 3))
#get input layer
input_layer = model.get_layer('input_1')
input_layer_in = input_layer.input
#get block5_pool layer
block5_pool = model.get_layer('block5_pool')
block5_pool_out = block5_pool.output
#get prediction layer
predictions_layer = model.get_layer('predictions')
predictions_layer_out = predictions_layer.output
#create new model using additional output of bottleneck features
vgg16_b_f_model = Model(inputs=input_layer_in,
outputs=[block5_pool_out, predictions_layer_out])
#view the summary of the new model
vgg16_b_f_model.summary()
#get input layer
b_f_input_layer = vgg16_b_f_model.get_layer('input_1')
b_f_input_layer_in = b_f_input_layer.input
print("The input layer: ", b_f_input_layer_in)
#get output layer
vgg16_b_f_outputs = vgg16_b_f_model.outputs
#convert input layer, block5 pool layer, and predictions layer from tensor to tensor info
input_layer_in_info = build_tensor_info(b_f_input_layer_in)
block5_pool_out_info = build_tensor_info(vgg16_b_f_outputs[0])
predictions_layer_out_info = build_tensor_info(vgg16_b_f_outputs[1])
print("The block5_pool layer: ", block5_pool_out_info)
print("The predictions layer: ", predictions_layer_out_info)
#export model
K.set_learning_phase(0)
export_path_base = "./vgg16_bottleneck-features"
model_version = "1"
export_path = os.path.join(tf.compat.as_bytes(export_path_base),
tf.compat.as_bytes(model_version))
print("Exporting trained model to: ", export_path)
prediction_signature = build_signature_def(
inputs={'images': input_layer_in_info},
outputs={
'scores_1': block5_pool_out_info,
'scores_2': predictions_layer_out_info
},
method_name=signature_constants.PREDICT_METHOD_NAME)
builder = saved_model_builder.SavedModelBuilder(export_path)
with K.get_session() as sess:
builder.add_meta_graph_and_variables(
sess=sess,
tags=[tag_constants.SERVING],
signature_def_map={'predict_images': prediction_signature})
builder.save()
#clear session
K.clear_session()
print("Done exporting!")
def _assertInvalidSignature(self, inputs, outputs, method_name):
signature_def = signature_def_utils_impl.build_signature_def(
inputs, outputs, method_name)
self.assertFalse(
signature_def_utils_impl.is_valid_signature(signature_def))
