diff --git a/mindspore/lite/tools/converter/parser/tf/tf_pad_parser.cc b/mindspore/lite/tools/converter/parser/tf/tf_pad_parser.cc
index 489a8c3603..04bc128f5c 100644
--- a/mindspore/lite/tools/converter/parser/tf/tf_pad_parser.cc
+++ b/mindspore/lite/tools/converter/parser/tf/tf_pad_parser.cc
@@ -31,7 +31,28 @@ ops::PrimitiveC *TFPadParser::Parse(const tensorflow::NodeDef &tf_op,
   if (tf_op.op() == "Pad") {
     prim->set_padding_mode(mindspore::PaddingMode::CONSTANT);
     prim->set_constant_value(0.0f);
-
+  } else if (tf_op.op() == "PadV2") {
+    prim->set_padding_mode(mindspore::PaddingMode::CONSTANT);
+    if (tf_op.input_size() < 3) {
+      MS_LOG(ERROR) << "tf padv2 input size less than 3, which is " << tf_op.input_size();
+      return nullptr;
+    }
+    auto &const_value_name = tf_op.input(2);
+    if (tf_node_map.find(const_value_name) == tf_node_map.end()) {
+      MS_LOG(ERROR) << "cannot find the input.";
+      return nullptr;
+    }
+    tensorflow::AttrValue attr_value;
+    if (!TensorFlowUtils::FindAttrValue(*tf_node_map.at(const_value_name), "value", &attr_value)) {
+      MS_LOG(ERROR) << "the input may be not const, which is not support now.";
+      return nullptr;
+    }
+    auto &tensor_proto = attr_value.tensor();
+    if (tensor_proto.dtype() != tensorflow::DT_FLOAT) {
+      MS_LOG(ERROR) << "input data type only support float now.";
+      return nullptr;
+    }
+    prim->set_constant_value(tensor_proto.float_val(0));
   } else if (tf_op.op() == "MirrorPad") {
     tensorflow::AttrValue attr_value;
     if (!TensorFlowUtils::FindAttrValue(tf_op, "mode", &attr_value)) {
@@ -58,6 +79,7 @@ ops::PrimitiveC *TFPadParser::Parse(const tensorflow::NodeDef &tf_op,
   return prim.release();
 }
 TFNodeRegistrar g_tfPadParser("Pad", new TFPadParser());
+TFNodeRegistrar g_tfPadV2Parser("PadV2", new TFPadParser());
 TFNodeRegistrar g_tfMirrorPadParser("MirrorPad", new TFPadParser());
 }  // namespace lite
 }  // namespace mindspore
diff --git a/mindspore/lite/tools/optimizer/graph/unify_format_pass.cc b/mindspore/lite/tools/optimizer/graph/unify_format_pass.cc
index 5d1dfafe7c..617e2c620f 100644
--- a/mindspore/lite/tools/optimizer/graph/unify_format_pass.cc
+++ b/mindspore/lite/tools/optimizer/graph/unify_format_pass.cc
@@ -807,7 +807,6 @@ bool UnifyFormatPass::BasicProcess(const FuncGraphPtr &func_graph, bool main_gra
       return false;
     }
   }
-  ResetSubGraphInput();
   return true;
 }
 
@@ -868,7 +867,6 @@ bool UnifyFormatPass::DecreaseTransposeForSingleOp(const FuncGraphPtr &func_grap
       return false;
     }
   }
-  ResetSubGraphInput();
   return true;
 }
 
@@ -1029,6 +1027,7 @@ bool UnifyFormatPass::RunOnlyForShape(const FuncGraphPtr &func_graph) {
     MS_LOG(ERROR) << "run framework transpose unify failed.";
     return false;
   }
+  ResetSubGraphInput();
   // delete insert transpose op and update op output shape.
   if (!ResetFuncGraph(func_graph)) {
     MS_LOG(ERROR) << "reset func_graph failed.";
@@ -1059,11 +1058,13 @@ bool UnifyFormatPass::Run(const FuncGraphPtr &func_graph) {
     MS_LOG(ERROR) << "run framework transpose unify failed.";
     return false;
   }
+  ResetSubGraphInput();
   // if input format of a certain op can be NHWC, can try transform this op to decrease the number of transpose op.
   if (!DecreaseTransposeForSingleOp(func_graph)) {
     MS_LOG(ERROR) << "run local trans insert optimizer failed.";
     return false;
   }
+  ResetSubGraphInput();
   // if input format of several ops surrounded only by transpose op all can be NHWC,
   // we can delete these transpose ops, and at the same time, transform these middle ops.
   if (!DecreaseTransposeForMultiOp(func_graph)) {
diff --git a/mindspore/lite/tools/optimizer/graph/unused_node_remove_pass.cc b/mindspore/lite/tools/optimizer/graph/unused_node_remove_pass.cc
index 8f0466bc03..3481597cc1 100644
--- a/mindspore/lite/tools/optimizer/graph/unused_node_remove_pass.cc
+++ b/mindspore/lite/tools/optimizer/graph/unused_node_remove_pass.cc
@@ -41,8 +41,8 @@ STATUS UnusedNodeRemovePass::ProcessGraph(const FuncGraphPtr &func_graph) {
         }
       }
     }
-    if (utils::isa<FuncGraphPtr>(node)) {
-      auto sub_graph = utils::cast<FuncGraphPtr>(node);
+    if (utils::isa<ValueNode>(node) && GetValueNode<FuncGraphPtr>(node) != nullptr) {
+      auto sub_graph = GetValueNode<FuncGraphPtr>(node);
       auto status = ProcessGraph(sub_graph);
       if (status != RET_OK) {
         MS_LOG(ERROR) << "process sub graph failed";
@@ -51,8 +51,10 @@ STATUS UnusedNodeRemovePass::ProcessGraph(const FuncGraphPtr &func_graph) {
     }
   }
   auto nodes = func_graph->nodes();
+  auto graph_inputs = func_graph->get_inputs();
   for (auto &node : nodes) {
-    if (vis.find(node) == vis.end()) {
+    if (vis.find(node) == vis.end() &&
+        std::find(graph_inputs.begin(), graph_inputs.end(), node) == graph_inputs.end()) {
       func_graph->DropNode(node);
     }
   }