How can I read in a remote image file in Tensorflow for classification?

Question

I trained a classifier in the Tensorflow for Poets tutorial .

Works great locally, but how can I modify this to use images when I have a URL?

   import tensorflow as tf

   # change this as you see fit
   image_path = sys.argv[1]

   # Read in the image_data
   image_data = tf.gfile.FastGFile(image_path, 'rb').read()

But how can I classify images from an external url? eg http://www.google.com/logo.png

Answer 1

Figured out we can use urllib.request.

Substitute:

   image_data = tf.gfile.FastGFile(image_path, 'rb').read()

With:

   # image_url = "http://www.google.com/logo.png"
   req = urllib.request.Request(image_url)
   response = urllib.request.urlopen(req)
   image_data = response.read()