diff --git a/README.md b/README.md
index a04fa86f8b8..03a61e5191f 100644
--- a/README.md
+++ b/README.md
@@ -50,6 +50,10 @@ You can use this as an example.
    
    (2). For Images, we can use [BBox-Label-Tool](https://github.com/puzzledqs/BBox-Label-Tool) to label objects.
 
+        The data I used for the demo was downloaded from [Google Images](https://images.google.com/), and hand-labeled by my intern employees. (Just kidding, I had to label it myself. Damn it...)
+  
+        Since I am training with only two classes, and that the signs have less distortions and variances (compared to person or car, for example), I only trained around 300 images for each class to get a decent performance. But if you are training with more classes or harder classes, I suggest you have at least 1000 images for each class.
+
 2. Create Annotation in Darknet Format 
    
    (1). If we choose to use VOC data to train, use [scripts/voc_label.py](https://github.com/Guanghan/darknet/blob/master/scripts/voc_label.py) to convert existing VOC annotations to darknet format.