diff --git a/robots-at-robots/research/dataset/armor_dataset_factory.py b/robots-at-robots/research/dataset/armor_dataset_factory.py
index 138de0d526bc80fa925aaaa7b01cefa07fb814e5..f56736ccb390855efd39979483b3970adaa1c0b8 100644
--- a/robots-at-robots/research/dataset/armor_dataset_factory.py
+++ b/robots-at-robots/research/dataset/armor_dataset_factory.py
@@ -31,6 +31,18 @@ class ArmorDatasetFactory:
 def save_digits_img():
     for j, (digit_img, armor, k, p) in enumerate(ArmorDatasetFactory.from_dataset(DJIROCODataset.Final)):
         # we ignore the 6s, because the format of their pictures is not accurate and we don't need them
+
+        # usage:
+        #       - remove the two if (line 40 and 53-54) to load all images
+        # or
+        #       - load bunch of images (not all at the same time, depending if you just want to do some tests)
+        #       1. start with j >= 0 at the top and j >= 500 at the bottom to load the 500 first images
+        #       2. increse these numbers as you load other images (to save time and not load images twice)
+
+
+        # watchout:
+        # some images have chinese symbols in their name, you will either need to change their name of find another workaround
+
         if armor.numero != 6 and j >= 5925:  # TODO change j value depending where you are in saving (to save time)
             # if j >= 50:
             #     break