[TF FE] Support unknown and dynamic rank Placeholder shape (#14211)

* [TF FE] Support unknown and undefined Placeholder shape

Signed-off-by: Kazantsev, Roman <roman.kazantsev@intel.com>

* Fix mistake in the test

Signed-off-by: Kazantsev, Roman <roman.kazantsev@intel.com>

src/frontends/tensorflow/src/decoder_proto.cpp

@@ -101,8 +101,11 @@ ov::Any DecoderProto::get_attribute(const std::string& name) const {
     case ::tensorflow::AttrValue::ValueCase::kI:
         return attrs[0].i();
     case ::tensorflow::AttrValue::ValueCase::kShape: {
-        std::vector<ov::Dimension> dims;
         const auto& tf_shape = attrs[0].shape();
+        if (tf_shape.unknown_rank()) {
+            return ov::PartialShape::dynamic();
+        }
+        std::vector<ov::Dimension> dims;
         for (int i = 0; i < tf_shape.dim_size(); i++) {
             dims.emplace_back(tf_shape.dim(i).size());
         }

src/frontends/tensorflow/src/input_model.cpp

@@ -115,8 +115,19 @@ void InputModel::InputModelTFImpl::loadPlaces() {
         m_op_places.push_back(op_place);
         m_op_places_map[op_name] = op_place;
         if (op_type == "Placeholder") {
-            auto pshape = node_decoder->get_attribute("shape").as<ov::PartialShape>();
+            auto pshape = ov::PartialShape::dynamic();
-            auto type = node_decoder->get_attribute("dtype").as<ov::element::Type>();
+            auto shape_any = node_decoder->get_attribute("shape");
+            if (shape_any.is<ov::PartialShape>()) {
+                // sometimes shape attribute can be absent in the graph
+                // so we need to check if Any object is initialized first
+                pshape = shape_any.as<ov::PartialShape>();
+            }
+            auto dtype_any = node_decoder->get_attribute("dtype");
+            auto placeholder_name = node_decoder->get_op_name();
+            FRONT_END_GENERAL_CHECK(
+                dtype_any.is<ov::element::Type>(),
+                "Incorrect input model: Placeholder node " + placeholder_name + " has unspecified type.");
+            auto type = dtype_any.as<ov::element::Type>();
             std::vector<std::string> names = {op_name};
             auto tensor_place = std::make_shared<TensorPlace>(m_input_model, pshape, type, names);
             m_tensor_places[op_name] = tensor_place;

src/frontends/tensorflow/tests/convert_tricky_models.cpp

@@ -0,0 +1,40 @@
+// Copyright (C) 2018-2022 Intel Corporation
+// SPDX-License-Identifier: Apache-2.0
+//
+
+#include <openvino/frontend/exception.hpp>
+#include <openvino/frontend/manager.hpp>
+#include <openvino/op/util/framework_node.hpp>
+
+#include "common_test_utils/ngraph_test_utils.hpp"
+#include "tf_utils.hpp"
+#include "utils.hpp"
+
+using namespace std;
+using namespace ngraph;
+using namespace ov::frontend;
+
+TEST(FrontEndConvertModelTest, test_undefined_input_shape) {
+    FrontEndManager fem;
+    FrontEnd::Ptr frontEnd;
+    InputModel::Ptr inputModel;
+    ASSERT_NO_THROW(frontEnd = fem.load_by_framework(TF_FE));
+    ASSERT_NE(frontEnd, nullptr);
+    auto model_filename = FrontEndTestUtils::make_model_path(string(TEST_TENSORFLOW_MODELS_DIRNAME) +
+                                                             string("undefined_input_shape/undefined_input_shape.pb"));
+    ASSERT_NO_THROW(inputModel = frontEnd->load(model_filename));
+    ASSERT_NE(inputModel, nullptr);
+    shared_ptr<ngraph::Function> function;
+    ASSERT_NO_THROW(function = frontEnd->convert(inputModel));
+    ASSERT_NE(function, nullptr);
+
+    for (auto& node : function->get_ordered_ops()) {
+        if (node->get_friendly_name() == "x") {
+            ASSERT_TRUE(node->get_output_partial_shape(0).same_scheme(ov::PartialShape::dynamic()));
+        } else if (node->get_friendly_name() == "y") {
+            ASSERT_TRUE(node->get_output_partial_shape(0).same_scheme(ov::PartialShape{2, 3}));
+        } else if (node->get_friendly_name() == "z") {
+            ASSERT_TRUE(node->get_output_partial_shape(0).same_scheme(ov::PartialShape::dynamic()));
+        }
+    }
+}

src/frontends/tensorflow/tests/test_models/models_pbtxt/undefined_input_shape.pbtxt

@@ -0,0 +1,75 @@
+node {
+  name: "x"
+  op: "Placeholder"
+  attr {
+    key: "dtype"
+    value {
+      type: DT_FLOAT
+    }
+  }
+  attr {
+    key: "shape"
+    value {
+      shape {
+        unknown_rank: true
+      }
+    }
+  }
+}
+node {
+  name: "y"
+  op: "Placeholder"
+  attr {
+    key: "dtype"
+    value {
+      type: DT_FLOAT
+    }
+  }
+  attr {
+    key: "shape"
+    value {
+      shape {
+        dim {
+          size: 2
+        }
+        dim {
+          size: 3
+        }
+      }
+    }
+  }
+}
+node {
+  name: "z"
+  op: "Placeholder"
+  attr {
+    key: "dtype"
+    value {
+      type: DT_FLOAT
+    }
+  }
+}
+node {
+  name: "add"
+  op: "AddV2"
+  input: "x"
+  input: "y"
+  attr {
+    key: "T"
+    value {
+      type: DT_FLOAT
+    }
+  }
+}
+node {
+  name: "Mul"
+  op: "Mul"
+  input: "add"
+  input: "z"
+  attr {
+    key: "T"
+    value {
+      type: DT_FLOAT
+    }
+  }
+}

src/frontends/tensorflow/tests/test_models/models_pbtxt/undefined_input_shape.py

@@ -0,0 +1,17 @@
+# Copyright (C) 2018-2022 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+
+
+import tensorflow.compat.v1 as tf
+
+tf.reset_default_graph()
+
+with tf.Session() as sess:
+    x = tf.placeholder(dtype=tf.float32, shape=None, name='x')
+    y = tf.placeholder(dtype=tf.float32, shape=[2, 3], name='y')
+    z = tf.placeholder(dtype=tf.float32, shape=None, name='z')
+    add = tf.add(x, y, name="add")
+    tf.multiply(add, z)
+
+    tf.global_variables_initializer()
+    tf.io.write_graph(sess.graph, '.', 'undefined_input_shape.pbtxt', as_text=True)