diff --git a/knn.py b/knn.py
new file mode 100644
index 0000000..f3f1e15
--- /dev/null
+++ b/knn.py
@@ -0,0 +1,26 @@
+import cv2
+import numpy as np
+
+# Now load the data
+with np.load('/home/varun/opencv/Trails/Test/knn_data.npz') as data:
+    print data.files
+    train = data['train']
+    train_labels = data['train_labels']
+
+imgread = cv2.imread('OutSample1 (1).png')
+img = cv2.resize(imgread,(150,75))
+gray = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
+
+test = gray.reshape(-1,11250).astype(np.float32)
+
+for i in range(1,7):
+    test_label = np.array([i])
+
+    knn = cv2.KNearest()
+    knn.train(train,train_labels)
+    ret,result,neighbours,dist = knn.find_nearest(test,k=5)
+
+    matches = result==test_label
+    correct = np.count_nonzero(matches)
+    accuracy = correct*100.0/result.size
+    print i,accuracy