Add ONNX model metadata to ov::Model (#13712)

* Add ONNX model metadata to ov::Model

* Add correct path to onnx models

src/frontends/onnx/frontend/src/core/graph.cpp

@@ -262,10 +262,21 @@ void Graph::remove_dangling_parameters() {
     }
 }
 
+void Graph::set_metadata(std::shared_ptr<ov::Model>& model) const {
+    const std::string framework_section = "framework";
+    const auto metadata = m_model->get_metadata();
+
+    for (const auto& pair : metadata) {
+        model->set_rt_info(pair.second, framework_section, pair.first);
+    }
+}
+
 std::shared_ptr<Function> Graph::convert() {
     convert_to_ngraph_nodes();
     remove_dangling_parameters();
-    return create_function();
+    auto function = create_function();
+    set_metadata(function);
+    return function;
 }
 
 OutputVector Graph::make_framework_nodes(const Node& onnx_node) {

src/frontends/onnx/frontend/src/core/graph.hpp

@@ -66,6 +66,7 @@ protected:
     void decode_to_framework_nodes();
     void convert_to_ngraph_nodes();
     void remove_dangling_parameters();
+    void set_metadata(std::shared_ptr<ov::Model>& model) const;
     std::shared_ptr<Function> create_function();
 
     ParameterVector m_parameters;

src/frontends/onnx/frontend/src/core/model.hpp

@@ -54,6 +54,16 @@ public:
         return m_model_proto->producer_version();
     }
 
+    std::map<std::string, std::string> get_metadata() const {
+        std::map<std::string, std::string> metadata;
+
+        const auto& model_metadata = m_model_proto->metadata_props();
+        for (const auto& prop : model_metadata) {
+            metadata.emplace(prop.key(), prop.value());
+        }
+        return metadata;
+    }
+
     /// \brief Access an operator object by its type name and domain name
     /// The function will return the operator object if it exists, or report an error
     /// in case of domain or operator absence.

src/frontends/onnx/tests/models/model_with_metadata.prototxt

@@ -0,0 +1,130 @@
+ir_version: 3
+producer_name: "nGraph ONNX Importer"
+graph {
+  node {
+    input: "A"
+    input: "B"
+    output: "out"
+    op_type: "PriorBoxClustered"
+    attribute {
+      name: "width"
+      floats: 0.10000000149011612
+      floats: 0.10000000149011612
+      floats: 0.20000000298023224
+      floats: 0.20000000298023224
+      type: FLOATS
+    }
+    attribute {
+      name: "height"
+      floats: 0.10000000149011612
+      floats: 0.10000000149011612
+      floats: 0.20000000298023224
+      floats: 0.20000000298023224
+      type: FLOATS
+    }
+    attribute {
+      name: "step_w"
+      f: 64
+      type: FLOAT
+    }
+    attribute {
+      name: "clip"
+      i: 1
+      type: INT
+    }
+    attribute {
+      name: "step_h"
+      f: 64
+      type: FLOAT
+    }
+    attribute {
+      name: "offset"
+      f: 0.5
+      type: FLOAT
+    }
+    attribute {
+      name: "variance"
+      floats: 0.10000000149011612
+      floats: 0.10000000149011612
+      floats: 0.20000000298023224
+      floats: 0.20000000298023224
+      type: FLOATS
+    }
+    domain: "org.openvinotoolkit"
+  }
+  name: "compute_graph"
+  input {
+    name: "A"
+    type {
+      tensor_type {
+        elem_type: 1
+        shape {
+          dim {
+            dim_value: 1
+          }
+          dim {
+            dim_value: 1
+          }
+          dim {
+            dim_value: 1
+          }
+          dim {
+            dim_value: 1
+          }
+        }
+      }
+    }
+  }
+  input {
+    name: "B"
+    type {
+      tensor_type {
+        elem_type: 1
+        shape {
+          dim {
+            dim_value: 1
+          }
+          dim {
+            dim_value: 1
+          }
+          dim {
+            dim_value: 1
+          }
+          dim {
+            dim_value: 1
+          }
+        }
+      }
+    }
+  }
+  output {
+    name: "out"
+    type {
+      tensor_type {
+        elem_type: 1
+        shape {
+          dim {
+            dim_value: 1
+          }
+          dim {
+            dim_value: 2
+          }
+          dim {
+            dim_value: 4
+          }
+        }
+      }
+    }
+  }
+}
+opset_import {
+  version: 1
+}
+metadata_props {
+  key: "meta_key1"
+  value: "meta_value1"
+}
+metadata_props {
+  key: "meta_key2"
+  value: "meta_value2"
+}

src/frontends/onnx/tests/onnx_importer_test.cpp

@@ -10,6 +10,7 @@
 
 #include "common_test_utils/file_utils.hpp"
 #include "onnx_import/onnx.hpp"
+#include "openvino/openvino.hpp"
 #include "openvino/util/file_util.hpp"
 
 TEST(ONNX_Importer_Tests, ImportBasicModel) {
@@ -159,3 +160,24 @@ TEST(ONNX_Importer_Tests, IsOperatorSupported) {
 
     ASSERT_TRUE(is_abs_op_supported);
 }
+
+TEST(ONNX_Importer_Tests, ImportModelWithoutMetadata) {
+    ov::Core core;
+    auto model = core.read_model(
+        CommonTestUtils::getModelFromTestModelZoo(ov::util::path_join({ONNX_MODELS_DIR, "priorbox_clustered.onnx"})));
+    ASSERT_FALSE(model->has_rt_info("framework"));
+}
+
+TEST(ONNX_Importer_Tests, ImportModelWithMetadata) {
+    ov::Core core;
+    auto model = core.read_model(
+        CommonTestUtils::getModelFromTestModelZoo(ov::util::path_join({ONNX_MODELS_DIR, "model_with_metadata.onnx"})));
+    ASSERT_TRUE(model->has_rt_info("framework"));
+
+    const auto rtinfo = model->get_rt_info();
+    auto metadata = rtinfo.at("framework").as<ov::AnyMap>();
+
+    ASSERT_EQ(metadata.size(), 2);
+    ASSERT_EQ(metadata["meta_key1"].as<std::string>(), "meta_value1");
+    ASSERT_EQ(metadata["meta_key2"].as<std::string>(), "meta_value2");
+}