diff --git a/README.md b/README.md
index 4f739a89abc6a56359a28b803eb5833b92d8e139..c85db582014fc0b42678a65c21fb221c5f72ac78 100644
--- a/README.md
+++ b/README.md
@@ -14,64 +14,14 @@ pip install git+https://github.com/PaddlePaddle/X2Paddle.git@develop
 ```
 
 ## How To Use
+### TensorFlow
 ```
 x2paddle --framework=tensorflow --model=tf_model.pb --save_dir=pd_model
 ```
-
-## 转换tensorflow vgg_16模型
-
-### 步骤一 下载模型参数文件
-```
-wget http://download.tensorflow.org/models/vgg_16_2016_08_28.tar.gz
-```
-
-### 步骤二 导出vgg_16的pb模型
-使用如下python脚本转换
-```
-import tensorflow.contrib.slim as slim
-from tensorflow.contrib.slim.nets import vgg
-from tensorflow.python.framework import graph_util
-import tensorflow as tf
-
-def freeze_model(sess, output_tensor_names, freeze_model_path):
-    out_graph = graph_util.convert_variables_to_constants(
-        sess, sess.graph.as_graph_def(), output_tensor_names)
-    with tf.gfile.GFile(freeze_model_path, 'wb') as f:
-        f.write(out_graph.SerializeToString())
-
-    print("freeze model saved in {}".format(freeze_model_path))
-
-with tf.Session() as sess:
-    inputs = tf.placeholder(dtype=tf.float32,
-                            shape=[None, 224, 224, 3],
-                            name="inputs")
-    logits, endpoint = vgg.vgg_16(inputs, num_classes=1000, is_training=False)
-    load_model = slim.assign_from_checkpoint_fn(
-        "vgg_16.ckpt", slim.get_model_variables("vgg_16"))
-    load_model(sess)
-
-    freeze_model(sess, ["vgg_16/fc8/squeezed"], "vgg16.pb")
+### Caffe
 ```
-
-### 步骤三 模型转换
-
+x2paddle --framework=caffe --proto=deploy.proto --weight=deploy.caffemodel --save_dir=pd_model
 ```
-x2paddle --framework=tensorflow \
-         --model=../vgg16.pb \
-         --save_dir=paddle_model
-```
-## 转换caffe SqueezeNet模型
 
-### 步骤一 下载模型参数文件和proto文件
-```
-wget https://github.com/DeepScale/SqueezeNet/blob/master/SqueezeNet_v1.1/squeezenet_v1.1.caffemodel
-wget https://github.com/DeepScale/SqueezeNet/blob/master/SqueezeNet_v1.1/deploy.prototxt
-```
-
-### 步骤二 模型转换
-
-```
-x2paddle --framework=caffe \
-         --weight=../squeezenet_v1.1.caffemodel \
-         --proto =../deploy.prototxt \
-         --save_dir=paddle_model
+## Related Docs
+[1. 如何导出TensorFlow的pb模型](export_tf_model.md)
diff --git a/export_tf_model.md b/export_tf_model.md
new file mode 100644
index 0000000000000000000000000000000000000000..d2f98b7446802e636841308fece28f394388c6c1
--- /dev/null
+++ b/export_tf_model.md
@@ -0,0 +1,44 @@
+## 如何导出TensorFlow模型
+
+本文档介绍如何将TensorFlow模型导出为X2Paddle支持的模型格式。
+
+TensorFlow提供了接口可将网络参数和网络结构同时保存到同一个文件中，并且只保存指定的前向计算子图，下面示例展示了如何导出tensorflow/models下的VGG16模型
+
+步骤一 下载模型参数文件
+```
+wget http://download.tensorflow.org/models/vgg_16_2016_08_28.tar.gz
+tar xzvf vgg_16_2016_08_28.tar.gz
+```
+
+步骤二 加载和导出模型
+```
+#coding: utf-8
+import tensorflow.contrib.slim as slim
+from tensorflow.contrib.slim.nets import vgg
+from tensorflow.python.framework import graph_util
+import tensorflow as tf
+
+# 固化模型函数
+# output_tensor_names: list，指定模型的输出tensor的name
+# freeze_model_path: 模型导出的文件路径
+def freeze_model(sess, output_tensor_names, freeze_model_path):
+    out_graph = graph_util.convert_variables_to_constants(
+        sess, sess.graph.as_graph_def(), output_tensor_names)
+    with tf.gfile.GFile(freeze_model_path, 'wb') as f:
+        f.write(out_graph.SerializeToString())
+
+    print("freeze model saved in {}".format(freeze_model_path))
+    
+# 加载模型参数
+sess = tf.Session()
+inputs = tf.placeholder(dtype=tf.float32,
+                        shape=[None, 224, 224, 3],
+                        name="inputs")
+logits, endpoint = vgg.vgg_16(inputs, num_classes=1000, is_training=False)
+load_model = slim.assign_from_checkpoint_fn(
+    "vgg_16.ckpt", slim.get_model_variables("vgg_16"))
+load_model(sess)
+
+# 导出模型
+freeze_model(sess, ["vgg_16/fc8/squeezed"], "vgg16.pb")
+```