diff --git a/tools/lite/benchmark.sh b/deploy/lite/benchmark/benchmark.sh
similarity index 100%
rename from tools/lite/benchmark.sh
rename to deploy/lite/benchmark/benchmark.sh
diff --git a/docs/en/extension/paddle_mobile_inference_en.md b/docs/en/extension/paddle_mobile_inference_en.md
index bf9252a01a796e5076f4e180c47f165f5f94ff27..44beba110a39160a5cc59bbbb6db4d91d4c01809 100644
--- a/docs/en/extension/paddle_mobile_inference_en.md
+++ b/docs/en/extension/paddle_mobile_inference_en.md
@@ -44,7 +44,7 @@ wget -c https://paddle-inference-dist.bj.bcebos.com/PaddleLite/benchmark_0/bench
 After the PC and mobile phone are successfully connected, use the following command to start the model evaluation.
 
 ```
-sh tools/lite/benchmark.sh ./benchmark_bin_v8 ./inference result_armv8.txt true
+sh deploy/lite/benchmark/benchmark.sh ./benchmark_bin_v8 ./inference result_armv8.txt true
 ```
 
 Where `./benchmark_bin_v8` is the path of the benchmark binary file, `./inference` is the path of all the models that need to be evaluated, `result_armv8.txt` is the result file, and the final parameter `true` means that the model will be optimized before evaluation. Eventually, the evaluation result file of `result_armv8.txt` will be saved in the current folder. The specific performances are as follows.
diff --git a/docs/en/tutorials/getting_started_en.md b/docs/en/tutorials/getting_started_en.md
index db780de21b61ca5e4f4ad2e03ca984d9542baa78..5cdfd409391922b7374dc84451f9b1f8a33c1162 100644
--- a/docs/en/tutorials/getting_started_en.md
+++ b/docs/en/tutorials/getting_started_en.md
@@ -226,13 +226,13 @@ Firstly, you should export inference model using `tools/export_model.py`.
 python tools/export_model.py \
     --model MobileNetV3_large_x1_0 \
     --pretrained_model ./output/MobileNetV3_large_x1_0/best_model/ppcls \
-    --output_path ./inference/cls_infer
+    --output_path ./inference
 ```
 
 Among them, the `--model` parameter is used to specify the model name, `--pretrained_model` parameter is used to specify the model file path, the path does not need to include the model file suffix name, and `--output_path` is used to specify the storage path of the converted model.
 
 **Note**:
-1. File prefix must be assigned in `--output_path`. If `--output_path=./inference/cls_infer`, then three files will be generated in the folder `inference`, they are `cls_infer.pdiparams`, `cls_infer.pdmodel` and `cls_infer.pdiparams.info`.
+1. If `--output_path=./inference`, then three files will be generated in the folder `inference`, they are `inference.pdiparams`, `inference.pdmodel` and `inference.pdiparams.info`.
 2. In the file `export_model.py:line53`, the `shape` parameter is the shape of the model input image, the default is `224*224`. Please modify it according to the actual situation, as shown below:
 
 ```python
@@ -243,20 +243,20 @@ Among them, the `--model` parameter is used to specify the model name, `--pretra
 54 ])
 ```
 
-The above command will generate the model structure file (`cls_infer.pdmodel`) and the model weight file (`cls_infer.pdiparams`), and then the inference engine can be used for inference:
+The above command will generate the model structure file (`inference.pdmodel`) and the model weight file (`inference.pdiparams`), and then the inference engine can be used for inference:
 
 ```bash
 python tools/infer/predict.py \
     --image_file image path \
-    --model_file "./inference/cls_infer.pdmodel" \
-    --params_file "./inference/cls_infer.pdiparams" \
+    --model_file "./inference/inference.pdmodel" \
+    --params_file "./inference/inference.pdiparams" \
     --use_gpu=True \
     --use_tensorrt=False
 ```
 Among them:
 + `image_file`: The path of the image file to be predicted, such as `./test.jpeg`;
-+ `model_file`: Model file path, such as `./MobileNetV3_large_x1_0/cls_infer.pdmodel`;
-+ `params_file`: Weight file path, such as `./MobileNetV3_large_x1_0/cls_infer.pdiparams`;
++ `model_file`: Model file path, such as `./MobileNetV3_large_x1_0/inference.pdmodel`;
++ `params_file`: Weight file path, such as `./MobileNetV3_large_x1_0/inference.pdiparams`;
 + `use_tensorrt`: Whether to use the TesorRT, default by `True`;
 + `use_gpu`: Whether to use the GPU, default by `True`
 + `enable_mkldnn`: Wheter to use `MKL-DNN`, default by `False`. When both `use_gpu` and `enable_mkldnn` are set to `True`, GPU is used to run and `enable_mkldnn` will be ignored.
diff --git a/docs/zh_CN/extension/paddle_mobile_inference.md b/docs/zh_CN/extension/paddle_mobile_inference.md
index 833231f736ffc19cb3b705025d7511dc8eba6bc4..529df90040b65ce23a5e024f9a07f54b51f68391 100644
--- a/docs/zh_CN/extension/paddle_mobile_inference.md
+++ b/docs/zh_CN/extension/paddle_mobile_inference.md
@@ -45,7 +45,7 @@ wget -c https://paddle-inference-dist.bj.bcebos.com/PaddleLite/benchmark_0/bench
 PC端和手机连接成功后，使用下面的命令开始模型评估。
 
 ```
-sh tools/lite/benchmark.sh ./benchmark_bin_v8 ./inference result_armv8.txt true
+sh deploy/lite/benchmark/benchmark.sh ./benchmark_bin_v8 ./inference result_armv8.txt true
 ```
 
 其中`./benchmark_bin_v8`为benchmark二进制文件路径，`./inference`为所有需要评测的模型的路径，`result_armv8.txt`为保存的结果文件，最后的参数`true`表示在评估之后会首先进行模型优化。最终在当前文件夹下会输出`result_armv8.txt`的评估结果文件，具体信息如下。
diff --git a/docs/zh_CN/tutorials/getting_started.md b/docs/zh_CN/tutorials/getting_started.md
index acbc53a083c120ff361b03fe031997bad11ae541..e3a7464e4d7adb8bf7d62af4e13c8bfe23187311 100644
--- a/docs/zh_CN/tutorials/getting_started.md
+++ b/docs/zh_CN/tutorials/getting_started.md
@@ -238,13 +238,13 @@ python tools/infer/infer.py \
 python tools/export_model.py \
     --model MobileNetV3_large_x1_0 \
     --pretrained_model ./output/MobileNetV3_large_x1_0/best_model/ppcls \
-    --output_path ./inference/cls_infer
+    --output_path ./inference
 ```
 
 其中，参数`--model`用于指定模型名称，`--pretrained_model`用于指定模型文件路径，该路径仍无需包含模型文件后缀名（如[1.3 模型恢复训练](#1.3)），`--output_path`用于指定转换后模型的存储路径。
 
 **注意**：
-1. `--output_path`中必须指定文件名的前缀，若`--output_path=./inference/cls_infer`，则会在`inference`文件夹下生成`cls_infer.pdiparams`、`cls_infer.pdmodel`和`cls_infer.pdiparams.info`文件。
+1. `--output_path`表示输出的inference模型文件夹路径，若`--output_path=./inference`，则会在`inference`文件夹下生成`inference.pdiparams`、`inference.pdmodel`和`inference.pdiparams.info`文件。
 2. 文件`export_model.py:line53`中，`shape`参数为模型输入图像的`shape`，默认为`224*224`，请根据实际情况修改，如下所示：
 ```python
 50 # Please modify the 'shape' according to actual needs
@@ -254,20 +254,20 @@ python tools/export_model.py \
 54 ])
 ```
 
-上述命令将生成模型结构文件（`cls_infer.pdmodel`）和模型权重文件（`cls_infer.pdiparams`），然后可以使用预测引擎进行推理：
+上述命令将生成模型结构文件（`inference.pdmodel`）和模型权重文件（`inference.pdiparams`），然后可以使用预测引擎进行推理：
 
 ```bash
 python tools/infer/predict.py \
     --image_file 图片路径 \
-    --model_file "./inference/cls_infer.pdmodel" \
-    --params_file "./inference/cls_infer.pdiparams" \
+    --model_file "./inference/inference.pdmodel" \
+    --params_file "./inference/inference.pdiparams" \
     --use_gpu=True \
     --use_tensorrt=False
 ```
 其中：
 + `image_file`：待预测的图片文件路径，如 `./test.jpeg`
-+ `model_file`：模型结构文件路径，如 `./inference/cls_infer.pdmodel`
-+ `params_file`：模型权重文件路径，如 `./inference/cls_infer.pdiparams`
++ `model_file`：模型结构文件路径，如 `./inference/inference.pdmodel`
++ `params_file`：模型权重文件路径，如 `./inference/inference.pdiparams`
 + `use_tensorrt`：是否使用 TesorRT 预测引擎，默认值：`True`
 + `use_gpu`：是否使用 GPU 预测，默认值：`True`
 + `enable_mkldnn`：是否启用`MKL-DNN`加速，默认为`False`。注意`enable_mkldnn`与`use_gpu`同时为`True`时，将忽略`enable_mkldnn`，而使用GPU运行。
diff --git a/tools/ema.py b/tools/ema.py
index 4f4596e1f08e6154bea11a86eb2a1fd31438d9f3..e41f472d6a1e448e76b922a718befbaa23d6ab51 100644
--- a/tools/ema.py
+++ b/tools/ema.py
@@ -1,3 +1,17 @@
+# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserve.
+#
+# 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 paddle
 import numpy as np
 
diff --git a/tools/export_model.py b/tools/export_model.py
index 5d2517fbccffe73229c37e982c4a495bb58a8d14..51b4fe2b6b8737fa2cd826b1708fc8c5d495cf79 100644
--- a/tools/export_model.py
+++ b/tools/export_model.py
@@ -33,8 +33,7 @@ def parse_args():
     parser = argparse.ArgumentParser()
     parser.add_argument("-m", "--model", type=str)
     parser.add_argument("-p", "--pretrained_model", type=str)
-    parser.add_argument(
-        "-o", "--output_path", type=str, default="./inference/cls_infer")
+    parser.add_argument("-o", "--output_path", type=str, default="./inference")
     parser.add_argument("--class_dim", type=int, default=1000)
     parser.add_argument("--load_static_weights", type=str2bool, default=False)
     parser.add_argument("--img_size", type=int, default=224)
@@ -73,7 +72,7 @@ def main():
             paddle.static.InputSpec(
                 shape=[None, 3, args.img_size, args.img_size], dtype='float32')
         ])
-    paddle.jit.save(model, args.output_path)
+    paddle.jit.save(model, os.path.join(args.output_path, "inference"))
 
 
 if __name__ == "__main__":
diff --git a/tools/infer/infer.py b/tools/infer/infer.py
index b4bf43e32671017412d250f64cc73d5d2a5dbd5d..67da0e31e267cf7f55d8531b577f56eafb2365e7 100644
--- a/tools/infer/infer.py
+++ b/tools/infer/infer.py
@@ -14,19 +14,21 @@
 
 import numpy as np
 import cv2
-import utils
 import shutil
 import os
 import sys
+
+import paddle
+import paddle.nn.functional as F
+
 __dir__ = os.path.dirname(os.path.abspath(__file__))
 sys.path.append(__dir__)
 sys.path.append(os.path.abspath(os.path.join(__dir__, '../..')))
 
 from ppcls.utils.save_load import load_dygraph_pretrain
 from ppcls.modeling import architectures
-
-import paddle
-import paddle.nn.functional as F
+import utils
+from utils import get_image_list
 
 
 def postprocess(outputs, topk=5):
@@ -36,23 +38,6 @@ def postprocess(outputs, topk=5):
     return zip(index, prob[index])
 
 
-def get_image_list(img_file):
-    imgs_lists = []
-    if img_file is None or not os.path.exists(img_file):
-        raise Exception("not found any img file in {}".format(img_file))
-
-    img_end = ['jpg', 'png', 'jpeg', 'JPEG', 'JPG', 'bmp']
-    if os.path.isfile(img_file) and img_file.split('.')[-1] in img_end:
-        imgs_lists.append(img_file)
-    elif os.path.isdir(img_file):
-        for single_file in os.listdir(img_file):
-            if single_file.split('.')[-1] in img_end:
-                imgs_lists.append(os.path.join(img_file, single_file))
-    if len(imgs_lists) == 0:
-        raise Exception("not found any img file in {}".format(img_file))
-    return imgs_lists
-
-
 def save_prelabel_results(class_id, input_filepath, output_idr):
     output_dir = os.path.join(output_idr, str(class_id))
     if not os.path.isdir(output_dir):
diff --git a/tools/infer/predict.py b/tools/infer/predict.py
index 885bb3379ee977355d7e421a19f24403cc6d4cf7..fdcbb6a942a0e09464a9f0953f6fe2814f000512 100644
--- a/tools/infer/predict.py
+++ b/tools/infer/predict.py
@@ -12,13 +12,15 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import sys
-sys.path.insert(0, ".")
-import tools.infer.utils as utils
 import numpy as np
 import cv2
 import time
 
+import sys
+sys.path.insert(0, ".")
+import tools.infer.utils as utils
+from tools.infer.utils import get_image_list
+
 
 def predict(args, predictor):
     input_names = predictor.get_input_names()
@@ -32,27 +34,33 @@ def predict(args, predictor):
     if not args.enable_benchmark:
         # for PaddleHubServing
         if args.hubserving:
-            img = args.image_file
+            img_list = [args.image_file]
         # for predict only
         else:
-            img = cv2.imread(args.image_file)[:, :, ::-1]
-        assert img is not None, "Error in loading image: {}".format(
-            args.image_file)
-        inputs = utils.preprocess(img, args)
-        inputs = np.expand_dims(
-            inputs, axis=0).repeat(
-                args.batch_size, axis=0).copy()
-        input_tensor.copy_from_cpu(inputs)
-
-        predictor.run()
-
-        output = output_tensor.copy_to_cpu()
-        classes, scores = utils.postprocess(output, args)
-        if args.hubserving:
-            return classes, scores
-        print("Current image file: {}".format(args.image_file))
-        print("\ttop-1 class: {0}".format(classes[0]))
-        print("\ttop-1 score: {0}".format(scores[0]))
+            img_list = get_image_list(args.image_file)
+
+        for idx, img_name in enumerate(img_list):
+            if not args.hubserving:
+                img = cv2.imread(img_name)[:, :, ::-1]
+                assert img is not None, "Error in loading image: {}".format(
+                    img_name)
+            else:
+                img = img_name
+            inputs = utils.preprocess(img, args)
+            inputs = np.expand_dims(
+                inputs, axis=0).repeat(
+                    args.batch_size, axis=0).copy()
+            input_tensor.copy_from_cpu(inputs)
+
+            predictor.run()
+
+            output = output_tensor.copy_to_cpu()
+            classes, scores = utils.postprocess(output, args)
+            if args.hubserving:
+                return classes, scores
+            print("Current image file: {}".format(img_name))
+            print("\ttop-1 class: {0}".format(classes[0]))
+            print("\ttop-1 score: {0}".format(scores[0]))
     else:
         for i in range(0, test_num + 10):
             inputs = np.random.rand(args.batch_size, 3, 224,
diff --git a/tools/infer/utils.py b/tools/infer/utils.py
index 9ea5e35422c57db49391a2e0551376068929e782..69fcd05a34faa0011902d29d6c94300b6dc0ee66 100644
--- a/tools/infer/utils.py
+++ b/tools/infer/utils.py
@@ -12,9 +12,11 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+import os
 import argparse
 import cv2
 import numpy as np
+
 from paddle.inference import Config
 from paddle.inference import create_predictor
 
@@ -125,6 +127,23 @@ def postprocess(output, args):
     return classes, scores
 
 
+def get_image_list(img_file):
+    imgs_lists = []
+    if img_file is None or not os.path.exists(img_file):
+        raise Exception("not found any img file in {}".format(img_file))
+
+    img_end = ['jpg', 'png', 'jpeg', 'JPEG', 'JPG', 'bmp']
+    if os.path.isfile(img_file) and img_file.split('.')[-1] in img_end:
+        imgs_lists.append(img_file)
+    elif os.path.isdir(img_file):
+        for single_file in os.listdir(img_file):
+            if single_file.split('.')[-1] in img_end:
+                imgs_lists.append(os.path.join(img_file, single_file))
+    if len(imgs_lists) == 0:
+        raise Exception("not found any img file in {}".format(img_file))
+    return imgs_lists
+
+
 class ResizeImage(object):
     def __init__(self, resize_short=None):
         self.resize_short = resize_short