diff --git a/model_zoo/official/cv/unet3d/README.md b/model_zoo/official/cv/unet3d/README.md
index a5cdac62ce7..5de7b472449 100644
--- a/model_zoo/official/cv/unet3d/README.md
+++ b/model_zoo/official/cv/unet3d/README.md
@@ -20,6 +20,10 @@
         - [Evaluation](#evaluation)
             - [Evaluating on Ascend](#training-on-ascend)
             - [Evaluating on GPU](#training-on-gpu)
+    - [Inference Process](#inference-process)
+        - [Export MindIR](#export-mindir)
+        - [Infer on Ascend310](#infer-on-ascend310)
+        - [result](#result)
     - [Model Description](#model-description)
         - [Performance](#performance)
             - [Evaluation Performance](#evaluation-performance)
@@ -348,6 +352,40 @@ eval average dice is 0.9502010010453671
 
 ```
 
+## Inference Process
+
+### [Export MindIR](#contents)
+
+```shell
+python export.py --ckpt_file [CKPT_PATH] --file_name [FILE_NAME] --file_format [FILE_FORMAT]
+```
+
+The ckpt_file parameter is required,
+`file_format` should be in ["AIR", "MINDIR"]
+
+### Infer on Ascend310
+
+Before performing inference, the mindir file must be exported by `export.py` script. We only provide an example of inference using MINDIR model.
+
+```shell
+# Ascend310 inference
+bash run_infer_310.sh [MINDIR_PATH] [NEED_PREPROCESS] [DEVICE_ID]
+```
+
+- `NEED_PREPROCESS` means weather need preprocess or not, it's value is 'y' or 'n'.
+- `DEVICE_ID` is optional, default value is 0.
+
+### result
+
+Inference result is saved in current path, you can find result like this in acc.log file.
+
+```shell
+
+# grep "eval average dice is:" acc.log
+eval average dice is 0.9502010010453671
+
+```
+
 ## [Model Description](#contents)
 
 ### [Performance](#contents)
diff --git a/model_zoo/official/cv/unet3d/ascend310_infer/CMakeLists.txt b/model_zoo/official/cv/unet3d/ascend310_infer/CMakeLists.txt
new file mode 100644
index 00000000000..ee3c8544734
--- /dev/null
+++ b/model_zoo/official/cv/unet3d/ascend310_infer/CMakeLists.txt
@@ -0,0 +1,14 @@
+cmake_minimum_required(VERSION 3.14.1)
+project(Ascend310Infer)
+add_compile_definitions(_GLIBCXX_USE_CXX11_ABI=0)
+set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -O0 -g -std=c++17 -Werror -Wall -fPIE -Wl,--allow-shlib-undefined")
+set(PROJECT_SRC_ROOT ${CMAKE_CURRENT_LIST_DIR}/)
+option(MINDSPORE_PATH "mindspore install path" "")
+include_directories(${MINDSPORE_PATH})
+include_directories(${MINDSPORE_PATH}/include)
+include_directories(${PROJECT_SRC_ROOT})
+find_library(MS_LIB libmindspore.so ${MINDSPORE_PATH}/lib)
+file(GLOB_RECURSE MD_LIB ${MINDSPORE_PATH}/_c_dataengine*)
+
+add_executable(main src/main.cc src/utils.cc)
+target_link_libraries(main ${MS_LIB} ${MD_LIB} gflags)
diff --git a/model_zoo/official/cv/unet3d/ascend310_infer/build.sh b/model_zoo/official/cv/unet3d/ascend310_infer/build.sh
new file mode 100644
index 00000000000..285514e19f2
--- /dev/null
+++ b/model_zoo/official/cv/unet3d/ascend310_infer/build.sh
@@ -0,0 +1,29 @@
+#!/bin/bash
+# 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.
+# ============================================================================
+if [ -d out ]; then
+    rm -rf out
+fi
+
+mkdir out
+cd out || exit
+
+if [ -f "Makefile" ]; then
+  make clean
+fi
+
+cmake .. \
+    -DMINDSPORE_PATH="`pip3.7 show mindspore-ascend | grep Location | awk '{print $2"/mindspore"}' | xargs realpath`"
+make
diff --git a/model_zoo/official/cv/unet3d/ascend310_infer/inc/utils.h b/model_zoo/official/cv/unet3d/ascend310_infer/inc/utils.h
new file mode 100644
index 00000000000..efebe03a8c1
--- /dev/null
+++ b/model_zoo/official/cv/unet3d/ascend310_infer/inc/utils.h
@@ -0,0 +1,32 @@
+/**
+ * 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_INFERENCE_UTILS_H_
+#define MINDSPORE_INFERENCE_UTILS_H_
+
+#include <sys/stat.h>
+#include <dirent.h>
+#include <vector>
+#include <string>
+#include <memory>
+#include "include/api/types.h"
+
+std::vector<std::string> GetAllFiles(std::string_view dirName);
+DIR *OpenDir(std::string_view dirName);
+std::string RealPath(std::string_view path);
+mindspore::MSTensor ReadFileToTensor(const std::string &file);
+int WriteResult(const std::string& imageFile, const std::vector<mindspore::MSTensor> &outputs);
+#endif
diff --git a/model_zoo/official/cv/unet3d/ascend310_infer/src/main.cc b/model_zoo/official/cv/unet3d/ascend310_infer/src/main.cc
new file mode 100644
index 00000000000..dffc0b56262
--- /dev/null
+++ b/model_zoo/official/cv/unet3d/ascend310_infer/src/main.cc
@@ -0,0 +1,129 @@
+/**
+ * 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 <sys/time.h>
+#include <gflags/gflags.h>
+#include <dirent.h>
+#include <iostream>
+#include <string>
+#include <algorithm>
+#include <iosfwd>
+#include <vector>
+#include <fstream>
+#include <sstream>
+
+#include "include/api/model.h"
+#include "include/api/context.h"
+#include "include/api/types.h"
+#include "include/api/serialization.h"
+#include "include/dataset/execute.h"
+#include "include/dataset/vision.h"
+#include "inc/utils.h"
+
+using mindspore::Context;
+using mindspore::Serialization;
+using mindspore::Model;
+using mindspore::Status;
+using mindspore::MSTensor;
+using mindspore::dataset::Execute;
+using mindspore::ModelType;
+using mindspore::GraphCell;
+using mindspore::kSuccess;
+
+DEFINE_string(mindir_path, "", "mindir path");
+DEFINE_string(input0_path, ".", "input0 path");
+DEFINE_int32(device_id, 0, "device id");
+
+int main(int argc, char **argv) {
+  gflags::ParseCommandLineFlags(&argc, &argv, true);
+  if (RealPath(FLAGS_mindir_path).empty()) {
+    std::cout << "Invalid mindir" << std::endl;
+    return 1;
+  }
+
+  auto context = std::make_shared<Context>();
+  auto ascend310 = std::make_shared<mindspore::Ascend310DeviceInfo>();
+  ascend310->SetDeviceID(FLAGS_device_id);
+  context->MutableDeviceInfo().push_back(ascend310);
+  mindspore::Graph graph;
+  Serialization::Load(FLAGS_mindir_path, ModelType::kMindIR, &graph);
+
+  Model model;
+  Status ret = model.Build(GraphCell(graph), context);
+  if (ret != kSuccess) {
+    std::cout << "ERROR: Build failed." << std::endl;
+    return 1;
+  }
+
+  std::vector<MSTensor> model_inputs = model.GetInputs();
+  if (model_inputs.empty()) {
+    std::cout << "Invalid model, inputs is empty." << std::endl;
+    return 1;
+  }
+
+  auto input0_files = GetAllFiles(FLAGS_input0_path);
+  if (input0_files.empty()) {
+    std::cout << "ERROR: input data empty." << std::endl;
+    return 1;
+  }
+
+  std::map<double, double> costTime_map;
+  size_t size = input0_files.size();
+
+  for (size_t i = 0; i < size; ++i) {
+    struct timeval start = {0};
+    struct timeval end = {0};
+    double startTimeMs;
+    double endTimeMs;
+    std::vector<MSTensor> inputs;
+    std::vector<MSTensor> outputs;
+    std::cout << "Start predict input files:" << input0_files[i] << std::endl;
+
+    auto input0 = ReadFileToTensor(input0_files[i]);
+    inputs.emplace_back(model_inputs[0].Name(), model_inputs[0].DataType(), model_inputs[0].Shape(),
+                        input0.Data().get(), input0.DataSize());
+
+    gettimeofday(&start, nullptr);
+    ret = model.Predict(inputs, &outputs);
+    gettimeofday(&end, nullptr);
+    if (ret != kSuccess) {
+      std::cout << "Predict " << input0_files[i] << " failed." << std::endl;
+      return 1;
+    }
+    startTimeMs = (1.0 * start.tv_sec * 1000000 + start.tv_usec) / 1000;
+    endTimeMs = (1.0 * end.tv_sec * 1000000 + end.tv_usec) / 1000;
+    costTime_map.insert(std::pair<double, double>(startTimeMs, endTimeMs));
+    WriteResult(input0_files[i], outputs);
+  }
+  double average = 0.0;
+  int inferCount = 0;
+
+  for (auto iter = costTime_map.begin(); iter != costTime_map.end(); iter++) {
+    double diff = 0.0;
+    diff = iter->second - iter->first;
+    average += diff;
+    inferCount++;
+  }
+  average = average / inferCount;
+  std::stringstream timeCost;
+  timeCost << "NN inference cost average time: "<< average << " ms of infer_count " << inferCount << std::endl;
+  std::cout << "NN inference cost average time: "<< average << "ms of infer_count " << inferCount << std::endl;
+  std::string fileName = "./time_Result" + std::string("/test_perform_static.txt");
+  std::ofstream fileStream(fileName.c_str(), std::ios::trunc);
+  fileStream << timeCost.str();
+  fileStream.close();
+  costTime_map.clear();
+  return 0;
+}
diff --git a/model_zoo/official/cv/unet3d/ascend310_infer/src/utils.cc b/model_zoo/official/cv/unet3d/ascend310_infer/src/utils.cc
new file mode 100644
index 00000000000..c947e4d5f45
--- /dev/null
+++ b/model_zoo/official/cv/unet3d/ascend310_infer/src/utils.cc
@@ -0,0 +1,129 @@
+/**
+ * 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 <fstream>
+#include <algorithm>
+#include <iostream>
+#include "inc/utils.h"
+
+using mindspore::MSTensor;
+using mindspore::DataType;
+
+std::vector<std::string> GetAllFiles(std::string_view dirName) {
+  struct dirent *filename;
+  DIR *dir = OpenDir(dirName);
+  if (dir == nullptr) {
+    return {};
+  }
+  std::vector<std::string> res;
+  while ((filename = readdir(dir)) != nullptr) {
+    std::string dName = std::string(filename->d_name);
+    if (dName == "." || dName == ".." || filename->d_type != DT_REG) {
+      continue;
+    }
+    res.emplace_back(std::string(dirName) + "/" + filename->d_name);
+  }
+  std::sort(res.begin(), res.end());
+  for (auto &f : res) {
+    std::cout << "image file: " << f << std::endl;
+  }
+  return res;
+}
+
+int WriteResult(const std::string& imageFile, const std::vector<MSTensor> &outputs) {
+  std::string homePath = "./result_Files";
+  for (size_t i = 0; i < outputs.size(); ++i) {
+    size_t outputSize;
+    std::shared_ptr<const void> netOutput;
+    netOutput = outputs[i].Data();
+    outputSize = outputs[i].DataSize();
+    int pos = imageFile.rfind('/');
+    std::string fileName(imageFile, pos + 1);
+    fileName.replace(fileName.find('.'), fileName.size() - fileName.find('.'), '_' + std::to_string(i) + ".bin");
+    std::string outFileName = homePath + "/" + fileName;
+    FILE * outputFile = fopen(outFileName.c_str(), "wb");
+    fwrite(netOutput.get(), outputSize, sizeof(char), outputFile);
+    fclose(outputFile);
+    outputFile = nullptr;
+  }
+  return 0;
+}
+
+mindspore::MSTensor ReadFileToTensor(const std::string &file) {
+  if (file.empty()) {
+    std::cout << "Pointer file is nullptr" << std::endl;
+    return mindspore::MSTensor();
+  }
+
+  std::ifstream ifs(file);
+  if (!ifs.good()) {
+    std::cout << "File: " << file << " is not exist" << std::endl;
+    return mindspore::MSTensor();
+  }
+
+  if (!ifs.is_open()) {
+    std::cout << "File: " << file << "open failed" << std::endl;
+    return mindspore::MSTensor();
+  }
+
+  ifs.seekg(0, std::ios::end);
+  size_t size = ifs.tellg();
+  mindspore::MSTensor buffer(file, mindspore::DataType::kNumberTypeUInt8, {static_cast<int64_t>(size)}, nullptr, size);
+
+  ifs.seekg(0, std::ios::beg);
+  ifs.read(reinterpret_cast<char *>(buffer.MutableData()), size);
+  ifs.close();
+
+  return buffer;
+}
+
+
+DIR *OpenDir(std::string_view dirName) {
+  if (dirName.empty()) {
+    std::cout << " dirName is null ! " << std::endl;
+    return nullptr;
+  }
+  std::string realPath = RealPath(dirName);
+  struct stat s;
+  lstat(realPath.c_str(), &s);
+  if (!S_ISDIR(s.st_mode)) {
+    std::cout << "dirName is not a valid directory !" << std::endl;
+    return nullptr;
+  }
+  DIR *dir;
+  dir = opendir(realPath.c_str());
+  if (dir == nullptr) {
+    std::cout << "Can not open dir " << dirName << std::endl;
+    return nullptr;
+  }
+  std::cout << "Successfully opened the dir " << dirName << std::endl;
+  return dir;
+}
+
+std::string RealPath(std::string_view path) {
+  char realPathMem[PATH_MAX] = {0};
+  char *realPathRet = nullptr;
+  realPathRet = realpath(path.data(), realPathMem);
+
+  if (realPathRet == nullptr) {
+    std::cout << "File: " << path << " is not exist.";
+    return "";
+  }
+
+  std::string realPath(realPathMem);
+  std::cout << path << " realpath is: " << realPath << std::endl;
+  return realPath;
+}
diff --git a/model_zoo/official/cv/unet3d/default_config.yaml b/model_zoo/official/cv/unet3d/default_config.yaml
index 9292e77eb8b..5e6c53fb9cd 100644
--- a/model_zoo/official/cv/unet3d/default_config.yaml
+++ b/model_zoo/official/cv/unet3d/default_config.yaml
@@ -13,7 +13,7 @@ output_path: "/cache/train"
 load_path: "/cache/checkpoint_path/"
 device_target: "Ascend"
 checkpoint_path: "./checkpoint/"
-checkpoint_file_path: "Unet3d-9-877.ckpt"
+checkpoint_file_path: "Unet3d-10-110.ckpt"
 
 # ==============================================================================
 # Training options
@@ -35,9 +35,13 @@ lower_limit: 3
 
 # Export options
 device_id: 0
-ckpt_file: ""
-file_name: ""
-file_format: ""
+ckpt_file: "./checkpoint/Unet3d-10-110.ckpt"
+file_name: "unet3d"
+file_format: "MINDIR"
+
+# 310 infer options
+pre_result_path: "./preprocess_Result"
+post_result_path: "./result_Files"
 
 ---
 # Help description for each configuration
diff --git a/model_zoo/official/cv/unet3d/export.py b/model_zoo/official/cv/unet3d/export.py
new file mode 100644
index 00000000000..c54fb4595c6
--- /dev/null
+++ b/model_zoo/official/cv/unet3d/export.py
@@ -0,0 +1,42 @@
+# 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.
+# ============================================================================
+
+import os
+import numpy as np
+from mindspore import dtype as mstype
+from mindspore import context, Tensor, export
+from mindspore.train.serialization import load_checkpoint, load_param_into_net
+from src.unet3d_model import UNet3d, UNet3d_
+from src.model_utils.config import config
+from src.model_utils.moxing_adapter import moxing_wrapper
+
+device_id = int(os.getenv('DEVICE_ID'))
+context.set_context(mode=context.GRAPH_MODE, device_target=config.device_target, save_graphs=False, device_id=device_id)
+
+@moxing_wrapper()
+def export_model(ckpt_path):
+    if config.device_target == 'Ascend':
+        network = UNet3d()
+    else:
+        network = UNet3d_()
+    network.set_train(False)
+    param_dict = load_checkpoint(ckpt_path)
+    load_param_into_net(network, param_dict)
+    image_shape = [config.batch_size, config.in_channels] + config.roi_size
+    window_image = Tensor(np.zeros(image_shape), mstype.float32)
+    export(network, window_image, file_name=config.file_name, file_format=config.file_format)
+
+if __name__ == '__main__':
+    export_model(ckpt_path=config.ckpt_file)
diff --git a/model_zoo/official/cv/unet3d/postprocess.py b/model_zoo/official/cv/unet3d/postprocess.py
new file mode 100644
index 00000000000..42398e5f906
--- /dev/null
+++ b/model_zoo/official/cv/unet3d/postprocess.py
@@ -0,0 +1,62 @@
+# 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.
+# ============================================================================
+
+import os
+import numpy as np
+from src.utils import create_sliding_window, CalculateDice
+from src.model_utils.config import config
+from src.model_utils.moxing_adapter import moxing_wrapper
+
+@moxing_wrapper()
+def cal_acc():
+    ''' calculate accuracy'''
+    index = 0
+    total_dice = 0
+    image_path = os.path.join(config.pre_result_path, "image")
+    seg_path = os.path.join(config.pre_result_path, "seg")
+    file_num = len(os.listdir(image_path))
+    for j in range(file_num):
+        image = np.load(os.path.join(image_path, "unet3d_bs" + str(config.batch_size) + "_" + str(j) + ".npy"))
+        seg = np.load(os.path.join(seg_path, "unet3d_bs" + str(config.batch_size) + "_" + str(j) + ".npy"))
+        _, slice_list = create_sliding_window(image, config.roi_size, config.overlap)
+        image_size = (config.batch_size, config.num_classes) + image.shape[2:]
+        output_image = np.zeros(image_size, np.float32)
+        count_map = np.zeros(image_size, np.float32)
+        importance_map = np.ones(config.roi_size, np.float32)
+        i = 0
+        w_name = "unet3d_bs" + str(config.batch_size) + "_bt" + str(j) + "_" + str(i) + "_0.bin"
+        w_path = os.path.join(config.post_result_path, w_name)
+        while os.path.isfile(w_path):
+            pred_shape = (config.batch_size, config.num_classes) + tuple(config.roi_size)
+            pred_probs = np.fromfile(w_path, np.float32).reshape(pred_shape)
+            slice_ = slice_list[i]
+            output_image[slice_] += pred_probs
+            count_map[slice_] += importance_map
+            i += 1
+            w_name = "unet3d_bs" + str(config.batch_size) + "_bt" + str(j) + "_" + str(i) + "_0.bin"
+            w_path = os.path.join(config.post_result_path, w_name)
+
+        output_image = output_image / count_map
+        dice, _ = CalculateDice(output_image, seg)
+        print("The {} batch dice is {}".format(index, dice), flush=True)
+        total_dice += dice
+        index = index + 1
+    eval_data_size = index
+    avg_dice = total_dice / eval_data_size
+    print("**********************End Eval***************************************")
+    print("eval average dice is {}".format(avg_dice))
+
+if __name__ == '__main__':
+    cal_acc()
diff --git a/model_zoo/official/cv/unet3d/preprocess.py b/model_zoo/official/cv/unet3d/preprocess.py
new file mode 100644
index 00000000000..cd48b058468
--- /dev/null
+++ b/model_zoo/official/cv/unet3d/preprocess.py
@@ -0,0 +1,58 @@
+# 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.
+# ============================================================================
+
+import os
+import numpy as np
+from src.dataset import create_dataset
+from src.utils import create_sliding_window
+from src.model_utils.config import config
+from src.model_utils.moxing_adapter import moxing_wrapper
+
+
+@moxing_wrapper()
+def gen_bin(data_path):
+    ''' generate bin files.'''
+    data_dir = data_path + "/image/"
+    seg_dir = data_path + "/seg/"
+    eval_dataset = create_dataset(data_path=data_dir, seg_path=seg_dir, is_training=False)
+    eval_data_size = eval_dataset.get_dataset_size()
+    print("train dataset length is:", eval_data_size)
+
+    window_path = os.path.join(config.pre_result_path, "00_data")
+    image_path = os.path.join(config.pre_result_path, "image")
+    seg_path = os.path.join(config.pre_result_path, "seg")
+    os.makedirs(window_path)
+    os.makedirs(image_path)
+    os.makedirs(seg_path)
+    j = 0
+
+    for batch in eval_dataset.create_dict_iterator(num_epochs=1, output_numpy=True):
+        image = batch["image"]
+        seg = batch["seg"]
+        f_name = "unet3d_bs" + str(config.batch_size) + "_" + str(j) + ".npy"
+        np.save(os.path.join(image_path, f_name), image)
+        np.save(os.path.join(seg_path, f_name), seg)
+        sliding_window_list, slice_list = create_sliding_window(image, config.roi_size, config.overlap)
+        i = 0
+        for window, _ in zip(sliding_window_list, slice_list):
+            w_name = "unet3d_bs" + str(config.batch_size) + "_bt" + str(j) + "_" + str(i) + ".bin"
+            window_file_path = os.path.join(window_path, w_name)
+            window.tofile(window_file_path)
+            i += 1
+        j += 1
+    print("=" * 20, "export bin files finished", "=" * 20)
+
+if __name__ == '__main__':
+    gen_bin(data_path=config.data_path)
diff --git a/model_zoo/official/cv/unet3d/scripts/run_infer_310.sh b/model_zoo/official/cv/unet3d/scripts/run_infer_310.sh
new file mode 100644
index 00000000000..713e516b5f5
--- /dev/null
+++ b/model_zoo/official/cv/unet3d/scripts/run_infer_310.sh
@@ -0,0 +1,120 @@
+#!/bin/bash
+# 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.
+# ============================================================================
+
+if [[ $# -lt 2 || $# -gt 3 ]]; then
+    echo "Usage: bash run_infer_310.sh [MINDIR_PATH] [NEED_PREPROCESS] [DEVICE_ID]
+    NEED_PREPROCESS means weather need preprocess or not, it's value is 'y' or 'n'.
+    DEVICE_ID is optional, it can be set by environment variable device_id, otherwise the value is zero"
+exit 1
+fi
+
+get_real_path(){
+    if [ "${1:0:1}" == "/" ]; then
+        echo "$1"
+    else
+        echo "$(realpath -m $PWD/$1)"
+    fi
+}
+model=$(get_real_path $1)
+
+if [ "$2" == "y" ] || [ "$2" == "n" ];then
+    need_preprocess=$2
+else
+  echo "weather need preprocess or not, it's value must be in [y, n]"
+  exit 1
+fi
+
+device_id=0
+if [ $# == 3 ]; then
+    device_id=$3
+fi
+
+echo "mindir name: "$model
+echo "need preprocess: "$need_preprocess
+echo "device id: "$device_id
+
+export ASCEND_HOME=/usr/local/Ascend/
+if [ -d ${ASCEND_HOME}/ascend-toolkit ]; then
+    export PATH=$ASCEND_HOME/fwkacllib/bin:$ASCEND_HOME/fwkacllib/ccec_compiler/bin:$ASCEND_HOME/ascend-toolkit/latest/fwkacllib/ccec_compiler/bin:$ASCEND_HOME/ascend-toolkit/latest/atc/bin:$PATH
+    export LD_LIBRARY_PATH=$ASCEND_HOME/fwkacllib/lib64:/usr/local/lib:$ASCEND_HOME/ascend-toolkit/latest/atc/lib64:$ASCEND_HOME/ascend-toolkit/latest/fwkacllib/lib64:$ASCEND_HOME/driver/lib64:$ASCEND_HOME/add-ons:$LD_LIBRARY_PATH
+    export TBE_IMPL_PATH=$ASCEND_HOME/ascend-toolkit/latest/opp/op_impl/built-in/ai_core/tbe
+    export PYTHONPATH=$ASCEND_HOME/fwkacllib/python/site-packages:${TBE_IMPL_PATH}:$ASCEND_HOME/ascend-toolkit/latest/fwkacllib/python/site-packages:$PYTHONPATH
+    export ASCEND_OPP_PATH=$ASCEND_HOME/ascend-toolkit/latest/opp
+else
+    export PATH=$ASCEND_HOME/fwkacllib/bin:$ASCEND_HOME/fwkacllib/ccec_compiler/bin:$ASCEND_HOME/atc/ccec_compiler/bin:$ASCEND_HOME/atc/bin:$PATH
+    export LD_LIBRARY_PATH=$ASCEND_HOME/fwkacllib/lib64:/usr/local/lib:$ASCEND_HOME/atc/lib64:$ASCEND_HOME/acllib/lib64:$ASCEND_HOME/driver/lib64:$ASCEND_HOME/add-ons:$LD_LIBRARY_PATH
+    export PYTHONPATH=$ASCEND_HOME/fwkacllib/python/site-packages:$ASCEND_HOME/atc/python/site-packages:$PYTHONPATH
+    export ASCEND_OPP_PATH=$ASCEND_HOME/opp
+fi
+
+function preprocess_data()
+{
+    if [ -d preprocess_Result ]; then
+        rm -rf ./preprocess_Result
+    fi
+    mkdir preprocess_Result
+    python3.7 ../preprocess.py
+}
+
+function compile_app()
+{
+    cd ../ascend310_infer || exit
+    bash build.sh &> build.log
+}
+
+function infer()
+{
+    cd - || exit
+    if [ -d result_Files ]; then
+        rm -rf ./result_Files
+    fi
+    if [ -d time_Result ]; then
+        rm -rf ./time_Result
+    fi
+    mkdir result_Files
+    mkdir time_Result
+
+    ../ascend310_infer/out/main --mindir_path=$model --input0_path=./preprocess_Result/00_data --device_id=$device_id &> infer.log
+
+}
+
+function cal_acc()
+{
+    python3.7 ../postprocess.py &> acc.log
+}
+
+if [ $need_preprocess == "y" ]; then
+    preprocess_data
+    if [ $? -ne 0 ]; then
+        echo "preprocess dataset failed"
+        exit 1
+    fi
+fi
+compile_app
+if [ $? -ne 0 ]; then
+    echo "compile app code failed"
+    exit 1
+fi
+infer
+if [ $? -ne 0 ]; then
+    echo " execute inference failed"
+    exit 1
+fi
+cal_acc
+if [ $? -ne 0 ]; then
+    echo "calculate accuracy failed"
+    exit 1
+fi
\ No newline at end of file