diff --git a/params.dat b/params.dat
new file mode 100644
index 0000000..cfb951a
--- /dev/null
+++ b/params.dat
@@ -0,0 +1,5 @@
+# parameters for watchdog
+model = /Users/mohamednouffer/workspace/akira_san/image_classifier/dataset/raw_data/arranged/mymodel2.h5
+monitering_folder = /Users/mohamednouffer/workspace/akira_san/image_classifier/dataset/raw_data/scanned/
+target_folder = /Users/mohamednouffer/workspace/akira_san/image_classifier/dataset/raw_data/categorised/
+image_size = 100
diff --git a/watcher.py b/watcher.py
new file mode 100644
index 0000000..4e93431
--- /dev/null
+++ b/watcher.py
@@ -0,0 +1,80 @@
+import sys
+import os, shutil
+import time
+import logging
+from watchdog.observers import Observer
+from watchdog.events import FileSystemEventHandler
+import numpy as np
+from tensorflow.keras.models import load_model
+from keras.preprocessing.image import load_img, ImageDataGenerator, array_to_img, img_to_array
+
+class fileMmonitor(FileSystemEventHandler):
+
+    def predict(self, model, image):
+        img_size = int(params['image_size'])
+        mdl = load_model(model)
+        target_folder = params['target_folder']
+        file_name = os.path.basename(image)
+        img = load_img(image, target_size=(img_size,img_size))
+        img = img_to_array(img)
+        img = img.reshape(1,img_size,img_size,3)
+        res = mdl.predict(img)
+        #print(res)
+        cat = np.argmax(res)
+        print(cat)
+        cat_path = os.path.join(target_folder,str(cat))
+        #dir_name = os.path.dirname(cat_path) 
+        if not os.path.exists(cat_path):
+            print("cat path created")
+            os.makedirs(cat_path)
+
+        try:
+            print(image)
+            print(os.path.join(cat_path, image))
+            shutil.move(image, os.path.join(cat_path, file_name))
+        except Exception as err:
+            print("An error was handled", err)
+        
+        #print([key for key in class_labels][np.argmax(res)])
+
+    def on_created(self, event):
+        #print(type(event))
+        print(event.src_path)
+        print(event.is_directory)
+        print(event.event_type)
+
+        fle = event.src_path
+        if fle.lower().endswith('jpg') or fle.lower().endswith('png') or fle.lower().endswith('jpeg'):
+            self.predict(params['model'], fle)
+        # #return super().on_created(event)
+        # for fle in os.listdir(folder_to_watch):
+        #     if fle.lower().endswith('jpg') or fle.lower().endswith('png') or fle.lower().endswith('jpeg'):
+        #         #print(fle);
+        #         pass;
+
+# read params from dat file
+fle = open("params.dat")
+params = {}
+for line in fle:
+    if line.startswith('#'):
+        continue
+    #print(line.strip())
+    key, val = line.split('=')
+    params[key.strip()] = val.strip()
+#print(params)
+
+print(params['monitering_folder'])    
+
+folder_to_watch = params['monitering_folder']
+observer = Observer()
+handler = fileMmonitor()
+observer.schedule(handler, folder_to_watch, recursive=True)
+observer.start()
+
+try:
+    while True:
+        time.sleep(2)
+except KeyboardInterrupt:
+    observer.stop()
+
+observer.join()