diff --git a/README.md b/README.md
index 502f9b2..24b6d3d 100644
--- a/README.md
+++ b/README.md
@@ -1,6 +1,6 @@
 
 
-<div align=center><img src="./picture/01.svg"/></div>
+<div align=center><img src="picture/01.svg?raw=true"/></div>
 # Edge-Engine
 
 
@@ -13,7 +13,7 @@
 
 email: zk@likedge.top
 
-<br>[![GitHub license](https://img.shields.io/badge/license-Apache--2.0-blue.svg)](https://raw.githubusercontent.com/tesseract-ocr/tesseract/master/LICENSE)</br>
+<br>[![GitHub license](https://img.shields.io/badge/license-Apache--2.0-blue.svg?raw=true)](https://raw.githubusercontent.com/tesseract-ocr/tesseract/master/LICENSE)</br>
 
 ------
 
@@ -36,7 +36,7 @@ email: zk@likedge.top
 > Target: x86_64-apple-darwin18.7.0
 > Thread model: posix
 
-<img src="image-20200418210521131.png" alt="image-20200418210521131"  />
+<img src="image-20200418210521131.png?raw=true" alt="image-20200418210521131"  />
 
 
 
@@ -44,7 +44,7 @@ email: zk@likedge.top
 
 这是什么?
 
-![path](path.png)
+![path](path.png?raw=true)
 
 ## 安装编译
 
@@ -101,7 +101,7 @@ edge_network作为一个类型存在,位于matrix_grad.h中结构体类型的数
 
 完整的序列模型:
 
-![image-20200128154352842](image-20200128154352842.png)
+![image-20200128154352842](image-20200128154352842.png?raw=true)
 
 ## 新的demo程序实现5层全连接层,可自定义神经元和激活函数,损失函数
 
@@ -130,7 +130,7 @@ Matrix output2 = sequaltial.forward(output1,weight2,bias2);
 
 同时第二层的输出也可以求出来,以此类推 .
 
-最终输出代码见nerual_test.cpp ![nerual_test1](/Users/zhangyiji/Documents/code/cpp_demo/my_os/Edge-Computing-Engine/picture/nerual_test1.png)
+最终输出代码见nerual_test.cpp ![nerual_test1](/Users/zhangyiji/Documents/code/cpp_demo/my_os/Edge-Computing-Engine/picture/nerual_test1.png?raw=true)
 
 代码:
 
@@ -467,7 +467,7 @@ function:
 
   Get:
 
-  ![](./picture/WX20191119-105411@2x.png)
+  ![](./picture/WX20191119-105411@2x.png?raw=true)