modify README

data/README.md

@@ -2,13 +2,13 @@
 
 We use ImageCLEF-DA, VisDA2017, Office-31 and Office-Home in our experiments.
 
-
-
 ## ImageCLEF-DA
 
-  
+We use ImageCLEF-DA published [here](https://github.com/thuml/Xlearn/tree/master/caffe), and replace the absolute path in `*List.txt` to relative path.
 
-```
+Please reorganize the images and label lists as following directory tree.
+
+```latex
 .
 |-- b
 |-- bList.txt
@@ -20,11 +20,13 @@ We use ImageCLEF-DA, VisDA2017, Office-31 and Office-Home in our experiments.
 `-- pList.txt
 ```
 
+## Office-31
 
+You can download the origin image [here](https://people.eecs.berkeley.edu/~jhoffman/domainadapt/). Our label files are replaced the absolute path  [here](https://github.com/thuml/PADA/tree/master/pytorch/data/office) to relative path.
 
-## Office-31
+Please reorganize the images and label lists as following directory tree.
 
-```
+```latex
 .
 |-- amazon
 |   |-- images
@@ -37,11 +39,13 @@ We use ImageCLEF-DA, VisDA2017, Office-31 and Office-Home in our experiments.
     `-- label.txt
 ```
 
+## Office-Home
 
+You can download the origin image [here](http://hemanthdv.org/OfficeHome-Dataset/). Our label files are replaced the absolute path  [here](https://github.com/thuml/PADA/tree/master/pytorch/data/office-home) to relative path.
 
-## Office-Home
+Please reorganize the images and label lists as following directory tree.
 
-```
+```latex
 .
 |-- Art
 |-- Art_shared.txt
@@ -57,11 +61,11 @@ We use ImageCLEF-DA, VisDA2017, Office-31 and Office-Home in our experiments.
 `-- Real_World.txt
 ```
 
-
-
 ## VisDA2017
 
-```
+You can download this dataset [here](https://github.com/VisionLearningGroup/taskcv-2017-public/tree/master/classification). We also announce the `validation6_list.txt` under the partial setting and label lists used in *sample size* experiment.
+
+```latex
 .
 |-- train
 |-- train6_list.txt