diff --git a/recognition/README.md b/recognition/README.md
index 77d49b2..e18dbcf 100644
--- a/recognition/README.md
+++ b/recognition/README.md
@@ -4,18 +4,21 @@ This part requires the following WARD 1.0 dataset:
 
     https://people.eecs.berkeley.edu/~yang/software/WAR/index.html
 
-However, you may also measure the similarity of any signals.
-For example, you may use it as:
-
-	   lambda = 0.1
-	   sigma = 0.2
-	   x = randn(10)
-	   y = randn(10)
-	   z = randn(10)
-
-       correlator = kcc_train(x, lambda, sigma);
-
-       similarity_xy = kcc(y, correlator);
-       similarity_xz = kcc(z, correlator);
-
-
+Unzip the databse to recognition folder, so that the directory looks like:
+
+    KCC
+    ├───recognition
+    │   ├───matrix2latex
+    │   ├───natsortfiles
+    │   │   └───html
+    │   └───WARD1.0
+    │       ├───Subject1
+    │       ├───Subject10
+    │       ├───Subject11
+    │       ├───Subject12
+    │       ├───Subject13
+    |       ...
+    └───tracking
+        └───utility
+
+Then you may find some examples in the run_me.m file