Merge "Get tensor infos from interpreter" into tizensubmit/tizen/20200602.011936

3 files changed, 56 insertions, 7 deletions

diff --git a/packaging/inference-engine-tflite.spec b/packaging/inference-engine-tflite.spec
index 9f2a99a..d9f9b72 100644
--- a/packaging/inference-engine-tflite.spec
+++ b/packaging/inference-engine-tflite.spec
@@ -1,7 +1,7 @@
 Name:       inference-engine-tflite
 Summary:    Tensorflow-Lite based implementation of inference-engine-interface
 Version:    0.0.1
-Release:    10
+Release:    11
 Group:      Multimedia/Libraries
 License:    Apache-2.0
 Source0:    %{name}-%{version}.tar.gz
diff --git a/src/inference_engine_tflite.cpp b/src/inference_engine_tflite.cpp
index bd44e06..11a7380 100644
--- a/src/inference_engine_tflite.cpp
+++ b/src/inference_engine_tflite.cpp
@@ -95,6 +95,7 @@ int InferenceTFLite::Load(std::vector<std::string> model_paths, inference_model_
 	}
 
     mInterpreter->SetNumThreads(MV_INFERENCE_TFLITE_MAX_THREAD_NUM);
+    LOGI("mInterpreter->tensors_size() :[%d]",mInterpreter->tensors_size());
 
     // input tensor
     if (mInterpreter->inputs().size()) {
@@ -151,8 +152,7 @@ int InferenceTFLite::GetInputTensorBuffers(std::vector<inference_engine_tensor_b
     LOGI("ENTER");
 
     if (mInputTensorInfo.empty()) {
-        LOGE("InputTensorInfo is empty. Do SetInputLayerProperty first.");
-        return INFERENCE_ENGINE_ERROR_INVALID_OPERATION;
+        SetInterpreterInfo();
     }
 
     mInputData.clear();
@@ -221,14 +221,12 @@ int InferenceTFLite::GetInputLayerProperty(inference_engine_layer_property &prop
 {
     LOGI("ENTER");
 
-    if (mInputLayer.empty()) {
-        return INFERENCE_ENGINE_ERROR_INVALID_OPERATION;
-    }
-
+    SetInterpreterInfo();
     property.layer_names = mInputLayer;
     property.tensor_infos = mInputTensorInfo;
 
     LOGI("LEAVE");
+
     return INFERENCE_ENGINE_ERROR_NONE;
 }
 
@@ -317,6 +315,7 @@ int InferenceTFLite::SetInputLayerProperty(inference_engine_layer_property &prop
 int InferenceTFLite::SetOutputLayerProperty(inference_engine_layer_property &property)
 {
     LOGI("ENTER");
+
     std::vector<std::string>::iterator iter;
     for (iter = property.layer_names.begin(); iter != property.layer_names.end(); iter++) {
         std::string name = *iter;
@@ -364,6 +363,54 @@ int InferenceTFLite::Run(std::vector<inference_engine_tensor_buffer> &input_buff
     return INFERENCE_ENGINE_ERROR_NONE;
 }
 
+int InferenceTFLite::SetInterpreterInfo()
+{
+    if (mInputLayer.empty() || mInputTensorInfo.empty()) {
+        LOGI("mInputLayer is empty. layers and tensors that mInterpreter has will be returned.");
+
+        mInputLayer.clear();
+        std::vector<std::string>().swap(mInputLayer);
+
+        mInputTensorInfo.clear();
+        std::vector<inference_engine_tensor_info>().swap(mInputTensorInfo);
+
+        for (auto iter = mInputLayerId.begin(); iter != mInputLayerId.end(); ++iter) {
+            mInputLayer.push_back(mInterpreter->tensor((*iter))->name);
+
+            std::vector<size_t> shape_nhwc;
+
+            for (int idx = 0; idx <mInterpreter->tensor((*iter))->dims->size; idx++) {
+                shape_nhwc.push_back(mInterpreter->tensor((*iter))->dims->data[idx]);
+            }
+
+            inference_engine_tensor_info tensor_info {
+                shape_nhwc, INFERENCE_TENSOR_SHAPE_NHWC, INFERENCE_TENSOR_DATA_TYPE_NONE, 1
+            };
+
+            if (mInterpreter->tensor((*iter))->type == kTfLiteUInt8) {
+                LOGI("type is kTfLiteUInt8");
+                tensor_info.data_type = INFERENCE_TENSOR_DATA_TYPE_UINT8;
+            }
+            else if (mInterpreter->tensor((*iter))->type == kTfLiteFloat32) {
+                LOGI("type is kTfLiteFloat32");
+                tensor_info.data_type = INFERENCE_TENSOR_DATA_TYPE_FLOAT32;
+            }
+            else {
+                LOGE("Not supported");
+                return INFERENCE_ENGINE_ERROR_NOT_SUPPORTED_FORMAT;
+            }
+
+            for (auto iter2 : tensor_info.shape)
+            {
+                tensor_info.size *= iter2;
+            }
+            mInputTensorInfo.push_back(tensor_info);
+        }
+    }
+
+    return INFERENCE_ENGINE_ERROR_NONE;
+}
+
 extern "C"
 {
 class IInferenceEngineCommon* EngineCommonInit(void)
diff --git a/src/inference_engine_tflite_private.h b/src/inference_engine_tflite_private.h
index 1ab36c1..bfe2d3b 100644
--- a/src/inference_engine_tflite_private.h
+++ b/src/inference_engine_tflite_private.h
@@ -71,6 +71,8 @@ public:
             std::vector<inference_engine_tensor_buffer> &output_buffers) override;
     
 private:
+    int SetInterpreterInfo();
+
     std::unique_ptr<tflite::Interpreter> mInterpreter;
     std::unique_ptr<tflite::FlatBufferModel> mFlatBuffModel;
     std::vector<void *> mInputData;