!7260 [MSLITE] Support assigned input data shapes while running models.

Merge pull request !7260 from wangshaocong/bugfix_master

mindspore/lite/tools/benchmark/benchmark.cc

@@ -406,6 +406,30 @@ int Benchmark::RunBenchmark() {
     std::cout << "CompileGraph failed while running ", model_name.c_str();
     return ret;
   }
+  if (!flags_->input_shape_list_.empty()) {
+    std::vector<std::vector<int>> input_shapes;
+    std::string input_dims_list = flags_->input_shape_list_;
+    while (!input_dims_list.empty()) {
+      auto position =
+        input_dims_list.find(";") != input_dims_list.npos ? input_dims_list.find(";") + 1 : input_dims_list.length();
+      std::string input_dims = input_dims_list.substr(0, position);
+      std::vector<int> input_shape;
+      while (!input_dims.empty()) {
+        auto pos = input_dims.find(",") != input_dims.npos ? input_dims.find(",") + 1 : input_dims.length();
+        std::string dim = input_dims.substr(0, pos);
+        input_shape.emplace_back(std::stoi(dim));
+        input_dims = input_dims.substr(pos);
+      }
+      input_shapes.emplace_back(input_shape);
+      input_dims_list = input_dims_list.substr(position);
+    }
+    ret = session_->Resize(session_->GetInputs(), input_shapes);
+    if (ret != RET_OK) {
+      MS_LOG(ERROR) << "Input tensor resize failed.";
+      std::cout << "Input tensor resize failed.";
+      return ret;
+    }
+  }
   model->Free();
   ms_inputs_ = session_->GetInputs();
   auto end_prepare_time = GetTimeUs();

mindspore/lite/tools/benchmark/benchmark.h

@@ -70,6 +70,8 @@ class MS_API BenchmarkFlags : public virtual FlagParser {
     AddFlag(&BenchmarkFlags::benchmark_data_type_, "benchmarkDataType",
             "Benchmark data type. FLOAT | INT32 | INT8 | UINT8", "FLOAT");
     AddFlag(&BenchmarkFlags::accuracy_threshold_, "accuracyThreshold", "Threshold of accuracy", 0.5);
+    AddFlag(&BenchmarkFlags::input_shape_list_, "inputShapes",
+            "Shape of input data, the format should be NHWC. e.g. 1,32,32,32;1,1,32,32,1", "");
   }
 
   ~BenchmarkFlags() override = default;
@@ -86,6 +88,7 @@ class MS_API BenchmarkFlags : public virtual FlagParser {
   InDataType in_data_type_;
   std::string in_data_type_in_ = "bin";
   int cpu_bind_mode_ = 1;
+  std::string input_shape_list_;
   // MarkPerformance
   int loop_count_;
   int num_threads_;

mindspore/lite/tools/converter/legacy_optimizer/graph/infershape_pass.cc

@@ -26,6 +26,9 @@ using mindspore::lite::Tensor;
 namespace mindspore {
 namespace lite {
 namespace {
+constexpr int DEFAULT_DIM_VALUE = -1;
+}
+namespace {
 std::vector<Tensor *> ConvertTensorToLiteTensor(MetaGraphT *graph, const std::vector<uint32_t> &tensor_indexs,
                                                 const schema::PrimitiveType node_type) {
   std::vector<Tensor *> lite_tensors;
@@ -85,6 +88,15 @@ void FreeTensors(std::vector<Tensor *> input_tensors, std::vector<Tensor *> outp
 }  // namespace
 STATUS InferShapePass::Run(MetaGraphT *graph) {
   MS_ASSERT(graph != nullptr);
+  for (auto idx : graph->inputIndex) {
+    auto input_tensor = graph->allTensors[idx].get();
+    for (auto &dim : input_tensor->dims) {
+      if (dim == 0) {
+        MS_LOG(WARNING) << "One dimension of the input shape is 0, which would be set to 32 as a default value.";
+        dim = DEFAULT_DIM_VALUE;
+      }
+    }
+  }
   for (auto iter = graph->nodes.begin(); iter != graph->nodes.end(); iter++) {
     auto &node = *iter;
     auto input_tensors = ConvertTensorToLiteTensor(graph, node->inputIndex, node->primitive->value.type);

mindspore/lite/tools/converter/parser/onnx/onnx_constant_of_shape_parser.cc

@@ -41,7 +41,14 @@ STATUS OnnxConstantOfShapeParser::Parse(const onnx::GraphProto &onnx_graph, cons
   for (const auto &onnx_node_attr : onnx_node.attribute()) {
     const auto &attribute_name = onnx_node_attr.name();
     if (attribute_name == "value") {
-      attr->value = static_cast<int32_t>(onnx_node_attr.i());
+      if (onnx_node_attr.type() == onnx::AttributeProto_AttributeType_TENSOR) {
+        auto tensor = onnx_node_attr.t();
+        if (tensor.data_type() == onnx::AttributeProto_AttributeType_FLOAT) {
+          attr->value = onnx_node_attr.f();
+        } else if (tensor.data_type() == onnx::AttributeProto_AttributeType_INT) {
+          attr->value = static_cast<int32_t>(onnx_node_attr.i());
+        }
+      }
     }
   }
 

mindspore/lite/tools/converter/parser/onnx/onnx_pool_parser.cc

@@ -66,14 +66,14 @@ STATUS OnnxPoolParser::Parse(const onnx::GraphProto &onnx_graph, const onnx::Nod
     const auto &attribute_name = onnx_node_attr.name();
     if (attribute_name == "kernel_shape") {
       if (onnx_node_attr.ints_size() == 2) {
-        attr->windowW = static_cast<int32_t>(onnx_node_attr.ints(0));
+        attr->windowH = static_cast<int32_t>(onnx_node_attr.ints(0));
-        attr->windowH = static_cast<int32_t>(onnx_node_attr.ints(1));
+        attr->windowW = static_cast<int32_t>(onnx_node_attr.ints(1));
       }
     }
     if (attribute_name == "strides") {
       if (onnx_node_attr.ints_size() == 2) {
-        attr->strideW = static_cast<int32_t>(onnx_node_attr.ints(0));
+        attr->strideH = static_cast<int32_t>(onnx_node_attr.ints(0));
-        attr->strideH = static_cast<int32_t>(onnx_node_attr.ints(1));
+        attr->strideW = static_cast<int32_t>(onnx_node_attr.ints(1));
       }
     }
     if (attribute_name == "auto_pad") {