diff --git a/mindspore/lite/src/runtime/kernel/arm/int8/arithmetic_int8.cc b/mindspore/lite/src/runtime/kernel/arm/int8/arithmetic_int8.cc
index 70363a6fda2..e105e2e3153 100644
--- a/mindspore/lite/src/runtime/kernel/arm/int8/arithmetic_int8.cc
+++ b/mindspore/lite/src/runtime/kernel/arm/int8/arithmetic_int8.cc
@@ -79,16 +79,19 @@ int ArithmeticInt8CPUKernel::Init() {
 
   auto *input0_tensor = in_tensors_.at(0);
   auto in0_quant_args = input0_tensor->quant_params();
+  CHECK_LESS_RETURN(in0_quant_args.size(), 1);
   quant_args_.in0_args_.scale_ = in0_quant_args.front().scale;
   quant_args_.in0_args_.zp_ = in0_quant_args.front().zeroPoint;
 
   auto *input1_tensor = in_tensors_.at(1);
   auto in1_quant_args = input1_tensor->quant_params();
+  CHECK_LESS_RETURN(in1_quant_args.size(), 1);
   quant_args_.in1_args_.scale_ = in1_quant_args.front().scale;
   quant_args_.in1_args_.zp_ = in1_quant_args.front().zeroPoint;
 
   auto *out_tensor = out_tensors_.at(kOutputIndex);
   auto out_quant_args = out_tensor->quant_params();
+  CHECK_LESS_RETURN(out_quant_args.size(), 1);
   quant_args_.out_args_.scale_ = out_quant_args.front().scale;
   quant_args_.out_args_.zp_ = out_quant_args.front().zeroPoint;
   if (!InferShapeDone()) {
diff --git a/mindspore/lite/src/runtime/kernel/arm/int8/arithmetic_self_int8.cc b/mindspore/lite/src/runtime/kernel/arm/int8/arithmetic_self_int8.cc
index 2a85a2b0e21..b2ea83fd73e 100644
--- a/mindspore/lite/src/runtime/kernel/arm/int8/arithmetic_self_int8.cc
+++ b/mindspore/lite/src/runtime/kernel/arm/int8/arithmetic_self_int8.cc
@@ -32,11 +32,13 @@ int ArithmeticSelfInt8CPUKernel::Init() {
   auto *input_tensor = in_tensors_.at(kInputIndex);
   CHECK_NULL_RETURN(input_tensor);
   auto in_quant_args = input_tensor->quant_params();
+  CHECK_LESS_RETURN(in_quant_args.size(), 1);
   para_->quant_arg_.in_args_.scale_ = in_quant_args.front().scale;
   para_->quant_arg_.in_args_.zp_ = in_quant_args.front().zeroPoint * (-1);
 
   auto *out_tensor = out_tensors_.at(kOutputIndex);
   auto out_quant_args = out_tensor->quant_params();
+  CHECK_LESS_RETURN(out_quant_args.size(), 1);
   para_->quant_arg_.out_args_.scale_ = out_quant_args.front().scale;
   para_->quant_arg_.out_args_.zp_ = out_quant_args.front().zeroPoint;
 
diff --git a/mindspore/lite/src/runtime/kernel/arm/int8/resize_int8.cc b/mindspore/lite/src/runtime/kernel/arm/int8/resize_int8.cc
index 3b3f4428ffe..0fcbf247ebf 100644
--- a/mindspore/lite/src/runtime/kernel/arm/int8/resize_int8.cc
+++ b/mindspore/lite/src/runtime/kernel/arm/int8/resize_int8.cc
@@ -94,9 +94,11 @@ int ResizeInt8CPUKernel::Init() {
     return RET_ERROR;
   }
   auto input = in_tensors_.at(0);
+  CHECK_LESS_RETURN(input->quant_params().size(), 1);
   quant_in_->zp_ = input->quant_params().front().zeroPoint;
   quant_in_->scale_ = input->quant_params().front().scale;
   auto output = out_tensors_.at(0);
+  CHECK_LESS_RETURN(output->quant_params().size(), 1);
   quant_out_->zp_ = output->quant_params().front().zeroPoint;
   quant_out_->scale_ = output->quant_params().front().scale;
 
diff --git a/mindspore/lite/src/runtime/kernel/arm/int8/scale_int8.cc b/mindspore/lite/src/runtime/kernel/arm/int8/scale_int8.cc
index 75d13bb1c14..671cc01f651 100644
--- a/mindspore/lite/src/runtime/kernel/arm/int8/scale_int8.cc
+++ b/mindspore/lite/src/runtime/kernel/arm/int8/scale_int8.cc
@@ -172,6 +172,9 @@ int ScaleInt8CPUKernel::InitQuantArgs() {
   auto input = in_tensors_.at(0);
   auto scale = in_tensors_.at(1);
   auto output = out_tensors_.at(0);
+  CHECK_LESS_RETURN(input->quant_params().size(), 1);
+  CHECK_LESS_RETURN(scale->quant_params().size(), 1);
+  CHECK_LESS_RETURN(output->quant_params().size(), 1);
   auto input_scale = input->quant_params().front().scale;
   auto scale_scale = scale->quant_params().front().scale;
   auto output_scale = output->quant_params().front().scale;
@@ -188,6 +191,7 @@ int ScaleInt8CPUKernel::InitQuantArgs() {
 
   if (in_tensors_.size() == kScaleBiasInputsSize) {
     auto offset = in_tensors_.at(kOffsetIndex);
+    CHECK_LESS_RETURN(offset->quant_params().size(), 1);
     auto offset_scale = offset->quant_params().front().scale;
     scale_param_->offset_zp_ = offset->quant_params().front().zeroPoint;