diff --git a/How to training your data on YOLO.md b/How to training your data on YOLO.md
new file mode 100644
index 00000000..08172d56
--- /dev/null
+++ b/How to training your data on YOLO.md	
@@ -0,0 +1,36 @@
+![Darknet Logo](http://pjreddie.com/media/files/darknet-black-small.png)
+
+#Darknet#
+Darknet is an open source neural network framework written in C and CUDA. It is fast, easy to install, and supports CPU and GPU computation.
+
+For more information see the [Darknet project website](http://pjreddie.com/darknet).
+
+For questions or issues please use the [Google Group](https://groups.google.com/forum/#!forum/darknet).
+
+
+----------
+
+#How to training your own data on YOLO#
+
+## **Steps:** ##
+1. first, prepare your data, including images and lables. Note the image extension name and label format.  
+    (1). The darknet support images: jpeg jpg. I had add some code to support other image format the same as OpenCV. Please see the [darknet pull 13](https://github.com/pjreddie/darknet/pull/13) or my github.  
+    (2). The label must be unix format. If you generate labels in Windows, you can use dos2unix tools ion Ubuntu to convert the format.  
+    (3). Please note the code in Data.c line 231：  
+
+	`char *labelpath = find_replace(path, "images", "images");`	
+
+	`labelpath = find_replace(labelpath, "JPEGImages", "labels");`	 
+	
+
+	**the darknet will replace images to images in trainging data's txt file.**
+
+2. Change the train.txt file's path in tolo.c line 17 18 and line 147：  
+	`char *train_images = "/data/voc/train.txt";`	
+
+    `char *backup_directory = "/home/pjreddie/backup/";`	
+
+	`list *plist = get_paths("/home/pjreddie/data/voc/2007_test.txt");`	
+	
+
+