diff --git a/data/nyud/README.md b/data/nyud/README.md
index 3bfe54d9638ec293939b3a93ffba3923da96254f..ea2f9df7fb61de13bd10a7388cfc82f09f9f4be1 100644
--- a/data/nyud/README.md
+++ b/data/nyud/README.md
@@ -7,7 +7,8 @@ While there are many labels, we follow the 40 class task defined by
 Saurabh Gupta, Pablo Arbelaez, and Jitendra Malik.
 CVPR 2013
 
-at http://www.cs.berkeley.edu/~sgupta/pdf/GuptaArbelaezMalikCVPR13.pdf.
+at http://www.cs.berkeley.edu/~sgupta/pdf/GuptaArbelaezMalikCVPR13.pdf .
+To reproduce the results of our paper, you must make use of the data from Gupta et al. at http://people.eecs.berkeley.edu/~sgupta/cvpr13/data.tgz .
 
 Refer to `classes.txt` for the listing of classes in model output order.
 Refer to `../nyud_layers.py` for the Python data layer for this dataset.