!7396 fix upsample opeartor of onnx to resize opeartor

Merge pull request !7396 from yankai10/merge_1016

mindspore/lite/src/ops/resize.cc

@@ -127,9 +127,60 @@ int Resize::InferShape(std::vector<lite::Tensor *> inputs_, std::vector<lite::Te
       return RET_INFER_INVALID;
     }
     size_t shape_size = shape_tensor->ElementsNum();
-    auto data = reinterpret_cast<int32_t *>(shape_tensor->data_c());
-    for (size_t i = 0; i < shape_size; i++) {
-      output_shape.push_back(data[i]);
+    switch (shape_size) {
+      case kDimension_4d: {
+        if (shape_tensor->data_type() == kNumberTypeInt32) {
+          auto data = reinterpret_cast<int32_t *>(shape_tensor->data_c());
+          if (data == nullptr) {
+            MS_LOG(INFO) << "Resize op size can't cast int.";
+            return RET_INFER_INVALID;
+          }
+          switch (shape_tensor->GetFormat()) {
+            case schema::Format_NCHW:
+              output_shape.push_back(data[2] * input->Height());
+              output_shape.push_back(data[3] * input->Width());
+              break;
+            case schema::Format_NHWC:
+              output_shape.push_back(data[1] * input->Height());
+              output_shape.push_back(data[2] * input->Width());
+              break;
+            default:
+              MS_LOG(INFO) << "Resize don't support tensor format.";
+              return RET_INFER_INVALID;
+          }
+        } else if (shape_tensor->data_type() == kNumberTypeFloat32) {
+          auto data = reinterpret_cast<float *>(shape_tensor->data_c());
+          if (data == nullptr) {
+            MS_LOG(INFO) << "Resize op size can't cast float.";
+            return RET_INFER_INVALID;
+          }
+          switch (shape_tensor->GetFormat()) {
+            case schema::Format_NCHW:
+              output_shape.push_back(data[2] * input->Height());
+              output_shape.push_back(data[3] * input->Width());
+              break;
+            case schema::Format_NHWC:
+              output_shape.push_back(data[1] * input->Height());
+              output_shape.push_back(data[2] * input->Width());
+              break;
+            default:
+              MS_LOG(INFO) << "Resize don't support tensor format.";
+              return RET_INFER_INVALID;
+          }
+        }
+        break;
+      }
+      default: {
+        auto data = reinterpret_cast<int32_t *>(shape_tensor->data_c());
+        if (data == nullptr) {
+          MS_LOG(INFO) << "Resize op size can't cast float.";
+          return RET_INFER_INVALID;
+        }
+        for (size_t i = 0; i < shape_size; i++) {
+          output_shape.push_back(data[i]);
+        }
+        break;
+      }
     }
   } else if (inputs_.size() == kSingleNum) {
     auto new_height = GetNewHeight();

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

@@ -32,7 +32,7 @@ STATUS OnnxUpsampleParser::Parse(const onnx::GraphProto &onnx_graph, const onnx:
     return RET_NULL_PTR;
   }
 
-  std::unique_ptr<schema::UpsampleT> attr = std::make_unique<schema::UpsampleT>();
+  std::unique_ptr<schema::ResizeT> attr = std::make_unique<schema::ResizeT>();
   if (attr == nullptr) {
     MS_LOG(ERROR) << "new op failed";
     return RET_NULL_PTR;
@@ -41,14 +41,19 @@ STATUS OnnxUpsampleParser::Parse(const onnx::GraphProto &onnx_graph, const onnx:
   for (const auto &onnx_node_attr : onnx_node.attribute()) {
     const auto &attribute_name = onnx_node_attr.name();
     if (attribute_name == "mode") {
-      attr->mode = onnx_node_attr.s();
-    } else if (attribute_name == "scales") {
-      for (int i = 0; i < onnx_node_attr.floats_size(); ++i) {
-        attr->scales[i] = onnx_node_attr.floats(i);
+      if ("nearest" == onnx_node_attr.s()) {
+        attr->method = schema::ResizeMethod_NEAREST_NEIGHBOR;
+      } else if ("bilinear" == onnx_node_attr.s()) {
+        attr->method = schema::ResizeMethod_BILINEAR;
+      } else {
+        MS_LOG(ERROR) << "Resize do not support upsample mode";
+        return RET_ERROR;
       }
     }
   }
-
+  attr->newWidth = 1;
+  attr->newHeight = 1;
+  attr->alignCorners = false;
   op->primitive->value.type = schema::PrimitiveType_Upsample;
   op->primitive->value.value = attr.release();
   return RET_OK;