diff --git a/mace/core/types.h b/mace/core/types.h
index aa1f9a89f911aaa076171bc8d7cc7322b767358c..5bdd4930c17aface45bf8859e3291e1d8464b228 100644
--- a/mace/core/types.h
+++ b/mace/core/types.h
@@ -64,6 +64,7 @@ MACE_MAPPING_DATA_TYPE_AND_ENUM(int32_t, DT_INT32);
 enum FrameworkType {
   TENSORFLOW = 0,
   CAFFE = 1,
+  ONNX = 2,
 };
 
 template <typename T>
diff --git a/mace/ops/deconv_2d.cc b/mace/ops/deconv_2d.cc
index 6453544ae92c75efc5560ef5f157dcbbfedb13d5..9eba1bd014c8c21e835a2b4d72664b6a19769951 100644
--- a/mace/ops/deconv_2d.cc
+++ b/mace/ops/deconv_2d.cc
@@ -66,12 +66,12 @@ class Deconv2dOp<DeviceType::CPU, float> : public Deconv2dOpBase {
     const Tensor *filter = this->Input(1);
     const Tensor *bias = nullptr;
     const Tensor *output_shape_tensor = nullptr;
-    if (model_type_ == CAFFE) {
-      bias = this->InputSize() >= 3 ? this->Input(2) : nullptr;
-    } else {
+    if (model_type_ == TENSORFLOW) {
       output_shape_tensor =
           this->InputSize() >= 3 ? this->Input(2) : nullptr;
       bias = this->InputSize() >= 4 ? this->Input(3) : nullptr;
+    } else {
+      bias = this->InputSize() >= 3 ? this->Input(2) : nullptr;
     }
     Tensor *output = this->Output(0);
 
@@ -182,13 +182,7 @@ class Deconv2dOp<DeviceType::GPU, float> : public Deconv2dOpBase {
         context, operator_def_.get(), 1,
         OpenCLBufferType::CONV2D_FILTER, mem_type)
                    == MaceStatus::MACE_SUCCESS);
-    if (model_type_ == FrameworkType::CAFFE) {
-      if (operator_def_->input_size() >= 3) {
-        MACE_CHECK(TransformFilter(
-            context, operator_def_.get(), 2,
-            OpenCLBufferType::ARGUMENT, mem_type) == MaceStatus::MACE_SUCCESS);
-      }
-    } else {
+    if (model_type_ == FrameworkType::TENSORFLOW) {
       if (operator_def_->input_size() >= 4) {
         MACE_CHECK(TransformFilter(
             context,
@@ -197,6 +191,12 @@ class Deconv2dOp<DeviceType::GPU, float> : public Deconv2dOpBase {
             OpenCLBufferType::ARGUMENT,
             mem_type) == MaceStatus::MACE_SUCCESS);
       }
+    } else {
+      if (operator_def_->input_size() >= 3) {
+        MACE_CHECK(TransformFilter(
+            context, operator_def_.get(), 2,
+            OpenCLBufferType::ARGUMENT, mem_type) == MaceStatus::MACE_SUCCESS);
+      }
     }
   }
   MaceStatus Run(OpContext *context) override {
@@ -204,12 +204,12 @@ class Deconv2dOp<DeviceType::GPU, float> : public Deconv2dOpBase {
     const Tensor *filter = this->Input(1);
     const Tensor *bias = nullptr;
     const Tensor *output_shape_tensor = nullptr;
-    if (model_type_ == CAFFE) {
-      bias = this->InputSize() >= 3 ? this->Input(2) : nullptr;
-    } else {
+    if (model_type_ == TENSORFLOW) {
       output_shape_tensor =
           this->InputSize() >= 3 ? this->Input(2) : nullptr;
       bias = this->InputSize() >= 4 ? this->Input(3) : nullptr;
+    } else {
+      bias = this->InputSize() >= 3 ? this->Input(2) : nullptr;
     }
     Tensor *output = this->Output(0);