diff --git a/mindspore/ccsrc/minddata/dataset/api/execute.cc b/mindspore/ccsrc/minddata/dataset/api/execute.cc
index a4dc135057..1e18a797e5 100644
--- a/mindspore/ccsrc/minddata/dataset/api/execute.cc
+++ b/mindspore/ccsrc/minddata/dataset/api/execute.cc
@@ -14,6 +14,8 @@
  * limitations under the License.
  */
 
+#include <algorithm>
+#include <fstream>
 #include "minddata/dataset/include/execute.h"
 #include "minddata/dataset/core/de_tensor.h"
 #include "minddata/dataset/core/device_resource.h"
@@ -30,15 +32,22 @@
 #endif
 #ifdef ENABLE_ACL
 #include "minddata/dataset/core/ascend_resource.h"
+#include "minddata/dataset/kernels/ir/vision/ascend_vision_ir.h"
 #endif
 
 namespace mindspore {
 namespace dataset {
 
+using json = nlohmann::json;
+struct Execute::ExtraInfo {
+  std::multimap<std::string, std::vector<uint32_t>> aipp_cfg_;
+};
+
 // FIXME - Temporarily overload Execute to support both TensorOperation and TensorTransform
 Execute::Execute(std::shared_ptr<TensorOperation> op, MapTargetDevice deviceType) {
   ops_.emplace_back(std::move(op));
   device_type_ = deviceType;
+  info_ = std::make_shared<ExtraInfo>();
 #ifdef ENABLE_ACL
   if (device_type_ == MapTargetDevice::kAscend310) {
     device_resource_ = std::make_shared<AscendResource>();
@@ -54,6 +63,7 @@ Execute::Execute(std::shared_ptr<TensorOperation> op, MapTargetDevice deviceType
 Execute::Execute(std::shared_ptr<TensorTransform> op, MapTargetDevice deviceType) {
   // Convert op from TensorTransform to TensorOperation
   std::shared_ptr<TensorOperation> operation;
+  info_ = std::make_shared<ExtraInfo>();
   if (deviceType == MapTargetDevice::kCpu) {
     operation = op->Parse();
   } else {
@@ -96,6 +106,7 @@ Execute::Execute(TensorTransform op, MapTargetDevice deviceType) {
 Execute::Execute(TensorTransform *op, MapTargetDevice deviceType) {
   // Convert op from TensorTransform to TensorOperation
   std::shared_ptr<TensorOperation> operation;
+  info_ = std::make_shared<ExtraInfo>();
   if (deviceType == MapTargetDevice::kCpu) {
     operation = op->Parse();
   } else {
@@ -117,6 +128,7 @@ Execute::Execute(TensorTransform *op, MapTargetDevice deviceType) {
 
 Execute::Execute(std::vector<std::shared_ptr<TensorOperation>> ops, MapTargetDevice deviceType)
     : ops_(std::move(ops)), device_type_(deviceType) {
+  info_ = std::make_shared<ExtraInfo>();
 #ifdef ENABLE_ACL
   if (device_type_ == MapTargetDevice::kAscend310) {
     device_resource_ = std::make_shared<AscendResource>();
@@ -131,6 +143,7 @@ Execute::Execute(std::vector<std::shared_ptr<TensorOperation>> ops, MapTargetDev
 
 Execute::Execute(std::vector<std::shared_ptr<TensorTransform>> ops, MapTargetDevice deviceType) {
   // Convert ops from TensorTransform to TensorOperation
+  info_ = std::make_shared<ExtraInfo>();
   if (deviceType == MapTargetDevice::kCpu) {
     (void)std::transform(ops.begin(), ops.end(), std::back_inserter(ops_),
                          [](std::shared_ptr<TensorTransform> operation) -> std::shared_ptr<TensorOperation> {
@@ -156,6 +169,7 @@ Execute::Execute(std::vector<std::shared_ptr<TensorTransform>> ops, MapTargetDev
 
 Execute::Execute(const std::vector<std::reference_wrapper<TensorTransform>> ops, MapTargetDevice deviceType) {
   // Convert ops from TensorTransform to TensorOperation
+  info_ = std::make_shared<ExtraInfo>();
   if (deviceType == MapTargetDevice::kCpu) {
     (void)std::transform(
       ops.begin(), ops.end(), std::back_inserter(ops_),
@@ -181,6 +195,7 @@ Execute::Execute(const std::vector<std::reference_wrapper<TensorTransform>> ops,
 // Execute function for the example vector case: auto decode(new vision::Decode());
 Execute::Execute(std::vector<TensorTransform *> ops, MapTargetDevice deviceType) {
   // Convert ops from TensorTransform to TensorOperation
+  info_ = std::make_shared<ExtraInfo>();
   if (deviceType == MapTargetDevice::kCpu) {
     (void)std::transform(
       ops.begin(), ops.end(), std::back_inserter(ops_),
@@ -268,7 +283,11 @@ Status Execute::operator()(const mindspore::MSTensor &input, mindspore::MSTensor
       device_input = std::move(device_output);
     }
     CHECK_FAIL_RETURN_UNEXPECTED(device_input->HasDeviceData(), "Apply transform failed, output tensor has no data");
-    *output = mindspore::MSTensor(std::make_shared<DETensor>(device_input, true));
+    std::shared_ptr<mindspore::dataset::Tensor> host_output;
+    // Need to optimize later, waiting for computing department development, hence we pop data temporarily.
+    RETURN_IF_NOT_OK(device_resource_->Pop(device_input, &host_output));
+    *output = mindspore::MSTensor(std::make_shared<DETensor>(host_output));
+    // *output = mindspore::MSTensor(std::make_shared<DETensor>(device_input, true)); Use in the future
 #endif
   }
   return Status::OK();
@@ -346,6 +365,169 @@ Status Execute::operator()(const std::vector<MSTensor> &input_tensor_list, std::
   return Status::OK();
 }
 
+std::vector<uint32_t> AippSizeFilter(const std::vector<uint32_t> &resize_para, const std::vector<uint32_t> &crop_para) {
+  std::vector<uint32_t> aipp_size;
+  if (resize_para.size() == 0) {
+    aipp_size = crop_para;
+  } else if (crop_para.size() == 0) {
+    aipp_size = resize_para;
+  } else {
+    if (resize_para.size() == 1) {
+      aipp_size = *min_element(crop_para.begin(), crop_para.end()) < *resize_para.begin() ? crop_para : resize_para;
+    } else {
+      aipp_size =
+        *min_element(resize_para.begin(), resize_para.end()) < *min_element(crop_para.begin(), crop_para.end())
+          ? resize_para
+          : crop_para;
+    }
+  }
+  return aipp_size;
+}
+
+std::vector<uint32_t> AippMeanFilter(const std::vector<uint32_t> &normalize_para) {
+  std::vector<uint32_t> aipp_mean;
+  if (normalize_para.size() == 6) {
+    std::transform(normalize_para.begin(), normalize_para.begin() + 3, std::back_inserter(aipp_mean),
+                   [](uint32_t i) { return static_cast<uint32_t>(i / 10000); });
+  } else {
+    aipp_mean = {0, 0, 0};
+  }
+  return aipp_mean;
+}
+
+std::vector<float> AippStdFilter(const std::vector<uint32_t> &normalize_para) {
+  std::vector<float> aipp_std;
+  if (normalize_para.size() == 6) {
+    auto zeros = std::find(std::begin(normalize_para), std::end(normalize_para), 0);
+    if (zeros == std::end(normalize_para)) {
+      std::transform(normalize_para.begin() + 3, normalize_para.end(), std::back_inserter(aipp_std),
+                     [](uint32_t i) { return static_cast<float>(10000 / i); });
+    } else {
+      MS_LOG(WARNING) << "Detect 0 in std vector, please verify your input";
+      aipp_std = {1.0, 1.0, 1.0};
+    }
+  } else {
+    aipp_std = {1.0, 1.0, 1.0};
+  }
+  return aipp_std;
+}
+
+Status AippInfoCollection(std::map<std::string, std::string> *aipp_options, const std::vector<uint32_t> &aipp_size,
+                          const std::vector<uint32_t> &aipp_mean, const std::vector<float> &aipp_std) {
+  aipp_options->insert(std::make_pair("related_input_rank", "0"));
+  aipp_options->insert(std::make_pair("src_image_size_w", std::to_string(aipp_size[1])));
+  aipp_options->insert(std::make_pair("src_image_size_h", std::to_string(aipp_size[1])));
+  aipp_options->insert(std::make_pair("crop", "false"));
+  aipp_options->insert(std::make_pair("input_format", "YUV420SP_U8"));
+  aipp_options->insert(std::make_pair("aipp_mode", "static"));
+  aipp_options->insert(std::make_pair("csc_switch", "true"));
+  aipp_options->insert(std::make_pair("rbuv_swap_switch", "false"));
+  std::vector<int32_t> color_space_matrix = {256, 0, 359, 256, -88, -183, 256, 454, 0};
+  int count = 0;
+  for (int i = 0; i < 3; i++) {
+    for (int j = 0; j < 3; j++) {
+      std::string key_word = "matrix_r" + std::to_string(i) + "c" + std::to_string(j);
+      aipp_options->insert(std::make_pair(key_word, std::to_string(color_space_matrix[count])));
+      ++count;
+    }
+  }
+  std::vector<uint32_t> color_space_bias = {0, 128, 128};
+  for (int i = 0; i < 3; i++) {
+    std::string key_word = "input_bias_" + std::to_string(i);
+    aipp_options->insert(std::make_pair(key_word, std::to_string(color_space_bias[i])));
+  }
+  for (int i = 0; i < aipp_mean.size(); i++) {
+    std::string key_word = "mean_chn_" + std::to_string(i);
+    aipp_options->insert(std::make_pair(key_word, std::to_string(aipp_mean[i])));
+  }
+  for (int i = 0; i < aipp_mean.size(); i++) {
+    std::string key_word = "min_chn_" + std::to_string(i);
+    aipp_options->insert(std::make_pair(key_word, "0.0"));
+  }
+  for (int i = 0; i < aipp_std.size(); i++) {
+    std::string key_word = "var_reci_chn_" + std::to_string(i);
+    aipp_options->insert(std::make_pair(key_word, std::to_string(aipp_std[i])));
+  }
+  return Status::OK();
+}
+
+std::string Execute::AippCfgGenerator() {
+  std::string config_location = "./aipp.cfg";
+#ifdef ENABLE_ACL
+  std::vector<uint32_t> paras;  // Record the parameters value of each Ascend operators
+  for (int32_t i = 0; i < ops_.size(); i++) {
+    json ir_info;
+    if (ops_[i] == nullptr) {
+      MS_LOG(ERROR) << "Input TensorOperation[" + std::to_string(i) + "] is null";
+      return "";
+    }
+    if (ops_[i]->ValidateParams() != Status::OK()) {
+      MS_LOG(ERROR) << "Input TensorOperation[" + std::to_string(i) + "] has wrong parameters";
+      return "";
+    }
+    ops_[i]->to_json(&ir_info);
+    std::multimap<std::string, std::string> op_list = {{vision::kDvppCropJpegOperation, "size"},
+                                                       {vision::kDvppDecodeResizeOperation, "size"},
+                                                       {vision::kDvppDecodeResizeCropOperation, "crop_size"},
+                                                       {vision::kDvppDecodeResizeCropOperation, "resize_size"},
+                                                       {vision::kDvppNormalizeOperation, "mean"},
+                                                       {vision::kDvppNormalizeOperation, "std"},
+                                                       {vision::kDvppResizeJpegOperation, "size"}};
+    for (auto pos = op_list.equal_range(ops_[i]->Name()); pos.first != pos.second; ++pos.first) {
+      auto paras_key_word = pos.first->second;
+      paras = ir_info[paras_key_word].get<std::vector<uint32_t>>();
+      info_->aipp_cfg_.insert(std::make_pair(ops_[i]->Name(), paras));
+    }
+  }
+  std::ofstream outfile;
+  outfile.open(config_location, std::ofstream::out);
+  if (!outfile.is_open()) {
+    MS_LOG(ERROR) << "Fail to open Aipp config file, please verify your system config(including authority)"
+                  << "We will return empty string which represent the location of Aipp config file in this case";
+    std::string except = "";
+    return except;
+  }
+  if (device_type_ == MapTargetDevice::kAscend310) {
+    // Process resize parameters and crop parameters to find out the final size of input data
+    std::vector<uint32_t> resize_paras;
+    std::vector<uint32_t> crop_paras;
+    auto iter = info_->aipp_cfg_.find(vision::kDvppResizeJpegOperation);
+    if (iter != info_->aipp_cfg_.end()) {
+      resize_paras = iter->second;
+    }
+    iter = info_->aipp_cfg_.find(vision::kDvppCropJpegOperation);
+    if (iter != info_->aipp_cfg_.end()) {
+      crop_paras = iter->second;
+      if (crop_paras.size() == 1) {
+        crop_paras.emplace_back(crop_paras[0]);
+      }
+    }
+    std::vector<uint32_t> aipp_size = AippSizeFilter(resize_paras, crop_paras);
+    // Process normalization parameters to find out the final normalization parameters for Aipp module
+    std::vector<uint32_t> normalize_paras;
+    if (info_->aipp_cfg_.find(vision::kDvppNormalizeOperation) != info_->aipp_cfg_.end()) {
+      for (auto pos = info_->aipp_cfg_.equal_range(vision::kDvppNormalizeOperation); pos.first != pos.second;
+           ++pos.first) {
+        auto mean_or_std = pos.first->second;
+        normalize_paras.insert(normalize_paras.end(), mean_or_std.begin(), mean_or_std.end());
+      }
+    }
+    std::vector<uint32_t> aipp_mean = AippMeanFilter(normalize_paras);
+    std::vector<float> aipp_std = AippStdFilter(normalize_paras);
+    std::map<std::string, std::string> aipp_options;
+    AippInfoCollection(&aipp_options, aipp_size, aipp_mean, aipp_std);
+    std::string tab_char(4, ' ');
+    outfile << "aipp_op {" << std::endl;
+    for (auto &option : aipp_options) {
+      outfile << tab_char << option.first << " : " << option.second << std::endl;
+    }
+    outfile << "}";
+    outfile.close();
+  }
+#endif
+  return config_location;
+}
+
 Status Execute::validate_device_() {
   if (device_type_ != MapTargetDevice::kCpu && device_type_ != MapTargetDevice::kAscend310) {
     std::string err_msg = "Your input device is not supported. (Option: CPU or Ascend310)";
diff --git a/mindspore/ccsrc/minddata/dataset/api/vision.cc b/mindspore/ccsrc/minddata/dataset/api/vision.cc
index b2c07bce30..5ae28268f0 100644
--- a/mindspore/ccsrc/minddata/dataset/api/vision.cc
+++ b/mindspore/ccsrc/minddata/dataset/api/vision.cc
@@ -191,6 +191,15 @@ Normalize::Normalize(std::vector<float> mean, std::vector<float> std) : mean_(me
 
 std::shared_ptr<TensorOperation> Normalize::Parse() { return std::make_shared<NormalizeOperation>(mean_, std_); }
 
+std::shared_ptr<TensorOperation> Normalize::Parse(const MapTargetDevice &env) {
+  if (env == MapTargetDevice::kAscend310) {
+#ifdef ENABLE_ACL
+    return std::make_shared<DvppNormalizeOperation>(mean_, std_);
+#endif
+  }
+  return std::make_shared<NormalizeOperation>(mean_, std_);
+}
+
 #ifndef ENABLE_ANDROID
 // NormalizePad Transform Operation.
 NormalizePad::NormalizePad(const std::vector<float> &mean, const std::vector<float> &std, const std::string &dtype)
diff --git a/mindspore/ccsrc/minddata/dataset/core/device_tensor.h b/mindspore/ccsrc/minddata/dataset/core/device_tensor.h
index bcebe4dfb4..8b9fe6c94e 100644
--- a/mindspore/ccsrc/minddata/dataset/core/device_tensor.h
+++ b/mindspore/ccsrc/minddata/dataset/core/device_tensor.h
@@ -53,8 +53,10 @@ class DeviceTensor : public Tensor {
   Status SetYuvStrideShape_(const uint32_t &width, const uint32_t &widthStride, const uint32_t &height,
                             const uint32_t &heightStride);
 
-  std::vector<uint32_t> YUV_shape_;
+  std::vector<uint32_t> YUV_shape_;  // YUV_shape_ = {width, widthStride, height, heightStride}
+
   uint8_t *device_data_;
+
   uint32_t size_;
 };
 
diff --git a/mindspore/ccsrc/minddata/dataset/include/execute.h b/mindspore/ccsrc/minddata/dataset/include/execute.h
index 4b8a85c869..ee1e8c8430 100644
--- a/mindspore/ccsrc/minddata/dataset/include/execute.h
+++ b/mindspore/ccsrc/minddata/dataset/include/execute.h
@@ -19,6 +19,7 @@
 
 #include <string>
 #include <vector>
+#include <map>
 #include <memory>
 #include "include/api/context.h"
 #include "include/api/types.h"
@@ -63,6 +64,8 @@ class Execute {
 
   Status DeviceMemoryRelease();
 
+  std::string AippCfgGenerator();
+
  private:
   Status validate_device_();
 
@@ -71,6 +74,9 @@ class Execute {
   MapTargetDevice device_type_;
 
   std::shared_ptr<DeviceResource> device_resource_;
+
+  struct ExtraInfo;
+  std::shared_ptr<ExtraInfo> info_;
 };
 
 }  // namespace dataset
diff --git a/mindspore/ccsrc/minddata/dataset/include/vision_lite.h b/mindspore/ccsrc/minddata/dataset/include/vision_lite.h
index 5660e7f2d3..ff82be7256 100644
--- a/mindspore/ccsrc/minddata/dataset/include/vision_lite.h
+++ b/mindspore/ccsrc/minddata/dataset/include/vision_lite.h
@@ -155,6 +155,8 @@ class Normalize : public TensorTransform {
   /// \return Shared pointer to TensorOperation object.
   std::shared_ptr<TensorOperation> Parse() override;
 
+  std::shared_ptr<TensorOperation> Parse(const MapTargetDevice &env) override;
+
  private:
   std::vector<float> mean_;
   std::vector<float> std_;
diff --git a/mindspore/ccsrc/minddata/dataset/kernels/image/dvpp/CMakeLists.txt b/mindspore/ccsrc/minddata/dataset/kernels/image/dvpp/CMakeLists.txt
index bd7696d63f..84d4274048 100644
--- a/mindspore/ccsrc/minddata/dataset/kernels/image/dvpp/CMakeLists.txt
+++ b/mindspore/ccsrc/minddata/dataset/kernels/image/dvpp/CMakeLists.txt
@@ -6,6 +6,7 @@ add_library(kernels-dvpp-image OBJECT
         dvpp_decode_resize_crop_jpeg_op.cc
         dvpp_decode_resize_jpeg_op.cc
         dvpp_decode_jpeg_op.cc
-        dvpp_resize_jpeg_op.cc
-        dvpp_decode_png_op.cc)
+        dvpp_decode_png_op.cc
+        dvpp_normalize_op.cc
+        dvpp_resize_jpeg_op.cc)
 add_dependencies(kernels-dvpp-image dvpp-utils)
\ No newline at end of file
diff --git a/mindspore/ccsrc/minddata/dataset/kernels/image/dvpp/dvpp_crop_jpeg_op.cc b/mindspore/ccsrc/minddata/dataset/kernels/image/dvpp/dvpp_crop_jpeg_op.cc
index 5fbf7ef854..1bd9635291 100644
--- a/mindspore/ccsrc/minddata/dataset/kernels/image/dvpp/dvpp_crop_jpeg_op.cc
+++ b/mindspore/ccsrc/minddata/dataset/kernels/image/dvpp/dvpp_crop_jpeg_op.cc
@@ -145,5 +145,6 @@ Status DvppCropJpegOp::SetAscendResource(const std::shared_ptr<DeviceResource> &
   processor_->SetCropParas(crop_width_, crop_height_);
   return Status::OK();
 }
+
 }  // namespace dataset
 }  // namespace mindspore
diff --git a/mindspore/ccsrc/minddata/dataset/kernels/image/dvpp/dvpp_decode_resize_jpeg_op.cc b/mindspore/ccsrc/minddata/dataset/kernels/image/dvpp/dvpp_decode_resize_jpeg_op.cc
index 2dc0af11e2..2cf8819bc2 100644
--- a/mindspore/ccsrc/minddata/dataset/kernels/image/dvpp/dvpp_decode_resize_jpeg_op.cc
+++ b/mindspore/ccsrc/minddata/dataset/kernels/image/dvpp/dvpp_decode_resize_jpeg_op.cc
@@ -134,5 +134,6 @@ Status DvppDecodeResizeJpegOp::SetAscendResource(const std::shared_ptr<DeviceRes
   processor_->SetResizeParas(resized_width_, resized_height_);
   return Status::OK();
 }
+
 }  // namespace dataset
 }  // namespace mindspore
diff --git a/mindspore/ccsrc/minddata/dataset/kernels/image/dvpp/dvpp_normalize_op.cc b/mindspore/ccsrc/minddata/dataset/kernels/image/dvpp/dvpp_normalize_op.cc
new file mode 100644
index 0000000000..ce6a9f4586
--- /dev/null
+++ b/mindspore/ccsrc/minddata/dataset/kernels/image/dvpp/dvpp_normalize_op.cc
@@ -0,0 +1,39 @@
+/**
+ * Copyright 2021 Huawei Technologies Co., Ltd
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#include <algorithm>
+#include "minddata/dataset/kernels/image/dvpp/dvpp_normalize_op.h"
+
+namespace mindspore {
+namespace dataset {
+Status DvppNormalizeOp::Compute(const std::shared_ptr<DeviceTensor> &input, std::shared_ptr<DeviceTensor> *output) {
+  const TensorShape dvpp_shape({1, 1, 1});
+  const DataType dvpp_data_type(DataType::DE_UINT8);
+  mindspore::dataset::DeviceTensor::CreateEmpty(dvpp_shape, dvpp_data_type, output);
+  std::vector<uint32_t> yuv_shape = input->GetYuvStrideShape();
+  (*output)->SetAttributes(input->GetDeviceBuffer(), input->DeviceDataSize(), yuv_shape[0], yuv_shape[1], yuv_shape[2],
+                           yuv_shape[3]);
+  if (!((*output)->HasDeviceData())) {
+    std::string error = "[ERROR] Fail to get the output result from device memory!";
+    RETURN_STATUS_UNEXPECTED(error);
+  }
+  return Status::OK();
+}
+
+Status DvppNormalizeOp::SetAscendResource(const std::shared_ptr<DeviceResource> &resource) { return Status::OK(); }
+
+}  // namespace dataset
+}  // namespace mindspore
diff --git a/mindspore/ccsrc/minddata/dataset/kernels/image/dvpp/dvpp_normalize_op.h b/mindspore/ccsrc/minddata/dataset/kernels/image/dvpp/dvpp_normalize_op.h
new file mode 100644
index 0000000000..7e4dbe09bb
--- /dev/null
+++ b/mindspore/ccsrc/minddata/dataset/kernels/image/dvpp/dvpp_normalize_op.h
@@ -0,0 +1,51 @@
+/**
+ * Copyright 2021 Huawei Technologies Co., Ltd
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+#ifndef MINDSPORE_CCSRC_MINDDATA_DATASET_KERNELS_IMAGE_DVPP_DVPP_NORMALIZE_JPEG_OP_H
+#define MINDSPORE_CCSRC_MINDDATA_DATASET_KERNELS_IMAGE_DVPP_DVPP_NORMALIZE_JPEG_OP_H
+
+#include <memory>
+#include <string>
+#include <vector>
+#include "minddata/dataset/core/device_tensor.h"
+#include "minddata/dataset/core/device_resource.h"
+#include "minddata/dataset/kernels/tensor_op.h"
+#include "minddata/dataset/util/status.h"
+#include "mindspore/core/utils/log_adapter.h"
+
+namespace mindspore {
+namespace dataset {
+class DvppNormalizeOp : public TensorOp {
+ public:
+  explicit DvppNormalizeOp(std::vector<float> mean, std::vector<float> std) : mean_(mean), std_(std) {}
+
+  ~DvppNormalizeOp() = default;
+
+  Status Compute(const std::shared_ptr<DeviceTensor> &input, std::shared_ptr<DeviceTensor> *output) override;
+
+  std::string Name() const override { return kDvppNormalizeOp; }
+
+  Status SetAscendResource(const std::shared_ptr<DeviceResource> &resource) override;
+
+ private:
+  std::vector<float> mean_;
+  std::vector<float> std_;
+};
+
+}  // namespace dataset
+}  // namespace mindspore
+
+#endif  // MINDSPORE_CCSRC_MINDDATA_DATASET_KERNELS_IMAGE_DVPP_DVPP_NORMALIZE_JPEG_OP_H
diff --git a/mindspore/ccsrc/minddata/dataset/kernels/ir/vision/ascend_vision_ir.cc b/mindspore/ccsrc/minddata/dataset/kernels/ir/vision/ascend_vision_ir.cc
index cd858a4f8f..12023063e7 100644
--- a/mindspore/ccsrc/minddata/dataset/kernels/ir/vision/ascend_vision_ir.cc
+++ b/mindspore/ccsrc/minddata/dataset/kernels/ir/vision/ascend_vision_ir.cc
@@ -26,6 +26,7 @@
 #include "minddata/dataset/kernels/image/dvpp/dvpp_decode_resize_crop_jpeg_op.h"
 #include "minddata/dataset/kernels/image/dvpp/dvpp_decode_jpeg_op.h"
 #include "minddata/dataset/kernels/image/dvpp/dvpp_decode_png_op.h"
+#include "minddata/dataset/kernels/image/dvpp/dvpp_normalize_op.h"
 #include "minddata/dataset/kernels/image/dvpp/dvpp_resize_jpeg_op.h"
 
 namespace mindspore {
@@ -241,6 +242,53 @@ Status DvppDecodePngOperation::ValidateParams() { return Status::OK(); }
 
 std::shared_ptr<TensorOp> DvppDecodePngOperation::Build() { return std::make_shared<DvppDecodePngOp>(); }
 
+// DvppNormalize
+DvppNormalizeOperation::DvppNormalizeOperation(const std::vector<float> &mean, const std::vector<float> &std)
+    : mean_(mean), std_(std) {}
+
+Status DvppNormalizeOperation::ValidateParams() {
+  if (mean_.size() != 3) {
+    std::string err_msg = "DvppNormalization:: mean expecting size 3, got size: " + std::to_string(mean_.size());
+    MS_LOG(ERROR) << err_msg;
+    RETURN_STATUS_SYNTAX_ERROR(err_msg);
+  }
+  if (std_.size() != 3) {
+    std::string err_msg = "DvppNormalization: std expecting size 3, got size: " + std::to_string(std_.size());
+    MS_LOG(ERROR) << err_msg;
+    RETURN_STATUS_SYNTAX_ERROR(err_msg);
+  }
+  if (*min_element(mean_.begin(), mean_.end()) < 0 || *max_element(mean_.begin(), mean_.end()) > 256) {
+    std::string err_msg =
+      "Normalization can take parameters in range [0, 256] according to math theory of mean and sigma, got mean "
+      "vector" +
+      std::to_string(std_.size());
+    MS_LOG(ERROR) << err_msg;
+    RETURN_STATUS_SYNTAX_ERROR(err_msg);
+  }
+  if (*min_element(std_.begin(), std_.end()) < 0 || *max_element(std_.begin(), std_.end()) > 256) {
+    std::string err_msg =
+      "Normalization can take parameters in range [0, 256] according to math theory of mean and sigma, got mean "
+      "vector" +
+      std::to_string(std_.size());
+    MS_LOG(ERROR) << err_msg;
+    RETURN_STATUS_SYNTAX_ERROR(err_msg);
+  }
+  return Status::OK();
+}
+
+std::shared_ptr<TensorOp> DvppNormalizeOperation::Build() {
+  std::shared_ptr<DvppNormalizeOp> tensor_op = std::make_shared<DvppNormalizeOp>(mean_, std_);
+  return tensor_op;
+}
+
+Status DvppNormalizeOperation::to_json(nlohmann::json *out_json) {
+  nlohmann::json args;
+  args["mean"] = mean_;
+  args["std"] = std_;
+  *out_json = args;
+  return Status::OK();
+}
+
 // DvppResizeOperation
 DvppResizeJpegOperation::DvppResizeJpegOperation(const std::vector<uint32_t> &resize) : resize_(resize) {}
 
diff --git a/mindspore/ccsrc/minddata/dataset/kernels/ir/vision/ascend_vision_ir.h b/mindspore/ccsrc/minddata/dataset/kernels/ir/vision/ascend_vision_ir.h
index 57ae9fd400..3148890845 100644
--- a/mindspore/ccsrc/minddata/dataset/kernels/ir/vision/ascend_vision_ir.h
+++ b/mindspore/ccsrc/minddata/dataset/kernels/ir/vision/ascend_vision_ir.h
@@ -40,6 +40,7 @@ constexpr char kDvppDecodeResizeOperation[] = "DvppDecodeResize";
 constexpr char kDvppDecodeResizeCropOperation[] = "DvppDecodeResizeCrop";
 constexpr char kDvppDecodeJpegOperation[] = "DvppDecodeJpeg";
 constexpr char kDvppDecodePngOperation[] = "DvppDecodePng";
+constexpr char kDvppNormalizeOperation[] = "DvppNormalize";
 constexpr char kDvppResizeJpegOperation[] = "DvppResizeJpeg";
 
 /* ####################################### Derived TensorOperation classes ################################# */
@@ -121,6 +122,25 @@ class DvppDecodePngOperation : public TensorOperation {
   std::string Name() const override { return kDvppDecodePngOperation; }
 };
 
+class DvppNormalizeOperation : public TensorOperation {
+ public:
+  explicit DvppNormalizeOperation(const std::vector<float> &mean, const std::vector<float> &std);
+
+  ~DvppNormalizeOperation() = default;
+
+  std::shared_ptr<TensorOp> Build() override;
+
+  Status ValidateParams() override;
+
+  std::string Name() const override { return kDvppNormalizeOperation; }
+
+  Status to_json(nlohmann::json *out_json) override;
+
+ private:
+  std::vector<float> mean_;
+  std::vector<float> std_;
+};
+
 class DvppResizeJpegOperation : public TensorOperation {
  public:
   explicit DvppResizeJpegOperation(const std::vector<uint32_t> &resize);
diff --git a/mindspore/ccsrc/minddata/dataset/kernels/tensor_op.cc b/mindspore/ccsrc/minddata/dataset/kernels/tensor_op.cc
index 3a2ed4f731..6b6d06b69e 100644
--- a/mindspore/ccsrc/minddata/dataset/kernels/tensor_op.cc
+++ b/mindspore/ccsrc/minddata/dataset/kernels/tensor_op.cc
@@ -75,5 +75,6 @@ Status TensorOp::SetAscendResource(const std::shared_ptr<DeviceResource> &resour
   return Status(StatusCode::kMDUnexpectedError,
                 "This is a CPU operator which doesn't have Ascend Resource. Please verify your context");
 }
+
 }  // namespace dataset
 }  // namespace mindspore
diff --git a/mindspore/ccsrc/minddata/dataset/kernels/tensor_op.h b/mindspore/ccsrc/minddata/dataset/kernels/tensor_op.h
index c5ad338694..6a5a173d6a 100644
--- a/mindspore/ccsrc/minddata/dataset/kernels/tensor_op.h
+++ b/mindspore/ccsrc/minddata/dataset/kernels/tensor_op.h
@@ -66,6 +66,7 @@ constexpr char kDvppDecodeResizeCropJpegOp[] = "DvppDecodeResizeCropJpegOp";
 constexpr char kDvppDecodeResizeJpegOp[] = "DvppDecodeResizeJpegOp";
 constexpr char kDvppDecodeJpegOp[] = "DvppDecodeJpegOp";
 constexpr char kDvppDecodePngOp[] = "DvppDecodePngOp";
+constexpr char kDvppNormalizeOp[] = "DvppNormalizeOp";
 constexpr char kDvppResizeJpegOp[] = "DvppResizeJpegOp";
 constexpr char kEqualizeOp[] = "EqualizeOp";
 constexpr char kHwcToChwOp[] = "HWC2CHWOp";
diff --git a/tests/st/cpp/dataset/test_de.cc b/tests/st/cpp/dataset/test_de.cc
index 32143ac83b..75ba9455fd 100644
--- a/tests/st/cpp/dataset/test_de.cc
+++ b/tests/st/cpp/dataset/test_de.cc
@@ -81,7 +81,7 @@ TEST_F(TestDE, TestDvpp) {
 
   // Check image info
   ASSERT_TRUE(rc.IsOk());
-  ASSERT_EQ(image.Shape().size(), 2);
+  ASSERT_EQ(image.Shape().size(), 3);
   int32_t real_h = 0;
   int32_t real_w = 0;
   int32_t remainder = crop_paras[crop_paras.size() - 1] % 16;
@@ -92,9 +92,15 @@ TEST_F(TestDE, TestDvpp) {
     real_h = (crop_paras[0] % 2 == 0) ? crop_paras[0] : crop_paras[0] + 1;
     real_w = (remainder == 0) ? crop_paras[1] : crop_paras[1] + 16 - remainder;
   }
+  /* Use in the future
   ASSERT_EQ(image.Shape()[0], real_h);  // For image in YUV format, each pixel takes 1.5 byte
   ASSERT_EQ(image.Shape()[1], real_w);
   ASSERT_EQ(image.DataSize(), real_h * real_w * 1.5);
+   */
+  ASSERT_EQ(image.Shape()[0], 1.5 * real_h * real_w);  // For image in YUV format, each pixel takes 1.5 byte
+  ASSERT_EQ(image.Shape()[1], 1);
+  ASSERT_EQ(image.Shape()[2], 1);
+  ASSERT_EQ(image.DataSize(), real_h * real_w * 1.5);
 #endif
 }
 
@@ -119,7 +125,7 @@ TEST_F(TestDE, TestDvppSinkMode) {
 
   // Check image info
   ASSERT_TRUE(rc.IsOk());
-  ASSERT_EQ(image.Shape().size(), 2);
+  ASSERT_EQ(image.Shape().size(), 3);
   int32_t real_h = 0;
   int32_t real_w = 0;
   int32_t remainder = crop_paras[crop_paras.size() - 1] % 16;
@@ -130,14 +136,15 @@ TEST_F(TestDE, TestDvppSinkMode) {
     real_h = (crop_paras[0] % 2 == 0) ? crop_paras[0] : crop_paras[0] + 1;
     real_w = (remainder == 0) ? crop_paras[1] : crop_paras[1] + 16 - remainder;
   }
-  ASSERT_EQ(image.Shape()[0], real_h);  // For image in YUV format, each pixel takes 1.5 byte
-  ASSERT_EQ(image.Shape()[1], real_w);
-  ASSERT_EQ(image.DataSize(), 1.5 * real_w * real_h);
+  ASSERT_EQ(image.Shape()[0], 1.5 * real_h * real_w);  // For image in YUV format, each pixel takes 1.5 byte
+  ASSERT_EQ(image.Shape()[1], 1);
+  ASSERT_EQ(image.Shape()[2], 1);
+  ASSERT_EQ(image.DataSize(), real_h * real_w * 1.5);
   Transform.DeviceMemoryRelease();
 #endif
 }
 
-TEST_F(TestDE, TestDvppDecodeResizeCrop) {
+TEST_F(TestDE, TestDvppDecodeResizeCropNormalize) {
 #ifdef ENABLE_ACL
   std::shared_ptr<mindspore::dataset::Tensor> de_tensor;
   mindspore::dataset::Tensor::CreateFromFile("./data/dataset/apple.jpg", &de_tensor);
@@ -146,18 +153,24 @@ TEST_F(TestDE, TestDvppDecodeResizeCrop) {
   // Define dvpp transform
   std::vector<int32_t> crop_paras = {416};
   std::vector<int32_t> resize_paras = {512};
+  std::vector<float> mean = {0.485 * 255, 0.456 * 255, 0.406 * 255};
+  std::vector<float> std = {0.229 * 255, 0.224 * 255, 0.225 * 255};
   auto decode(new vision::Decode());
   auto resize(new vision::Resize(resize_paras));
   auto centercrop(new vision::CenterCrop(crop_paras));
-  std::vector<TensorTransform *> transforms = {decode, resize, centercrop};
-  mindspore::dataset::Execute Transform(transforms, MapTargetDevice::kAscend310);
+  auto normalize(new vision::Normalize(mean, std));
+  std::vector<TensorTransform *> trans_lists = {decode, resize, centercrop, normalize};
+  mindspore::dataset::Execute Transform(trans_lists, MapTargetDevice::kAscend310);
+
+  std::string aipp_cfg = Transform.AippCfgGenerator();
+  ASSERT_EQ(aipp_cfg, "./aipp.cfg");
 
   // Apply transform on images
   Status rc = Transform(image, &image);
 
   // Check image info
   ASSERT_TRUE(rc.IsOk());
-  ASSERT_EQ(image.Shape().size(), 2);
+  ASSERT_EQ(image.Shape().size(), 3);
   int32_t real_h = 0;
   int32_t real_w = 0;
   int32_t remainder = crop_paras[crop_paras.size() - 1] % 16;
@@ -168,9 +181,10 @@ TEST_F(TestDE, TestDvppDecodeResizeCrop) {
     real_h = (crop_paras[0] % 2 == 0) ? crop_paras[0] : crop_paras[0] + 1;
     real_w = (remainder == 0) ? crop_paras[1] : crop_paras[1] + 16 - remainder;
   }
-  ASSERT_EQ(image.Shape()[0], real_h);  // For image in YUV format, each pixel takes 1.5 byte
-  ASSERT_EQ(image.Shape()[1], real_w);
-  ASSERT_EQ(image.DataSize(), 1.5 * real_w * real_h);
+  ASSERT_EQ(image.Shape()[0], 1.5 * real_h * real_w);  // For image in YUV format, each pixel takes 1.5 byte
+  ASSERT_EQ(image.Shape()[1], 1);
+  ASSERT_EQ(image.Shape()[2], 1);
+  ASSERT_EQ(image.DataSize(), real_h * real_w * 1.5);
   Transform.DeviceMemoryRelease();
 #endif
 }