index.py

import tensorflow as tf
import requests
import responder
import os, tempfile

# download("http://gajamed.pl/uploads/2016/04/29.jpg")
# => "/tmp/foo.jpg"
def download(url):
    dest = os.path.join(tempfile.mkdtemp(), 'image.jpg')
    response = requests.get(url, allow_redirects=True)
    with open(dest, 'wb') as f:
        f.write(response.content)
    return dest

# load_graph("output_graph.pb")
# => tensorflow graph
def load_graph(filename):
    graph = tf.Graph()
    graph_def = tf.GraphDef()

    with open(filename, "rb") as f:
        graph_def.ParseFromString(f.read())
    with graph.as_default():
        tf.import_graph_def(graph_def)

    return graph

# read_file("/tmp/foo.jpg")
# => tensorflow representation of image (299 x 299 px)
def read_image(file_name):
    height, width = [299, 299]
    file_reader = tf.read_file(file_name, "file_reader")
    image_reader = tf.image.decode_jpeg(file_reader, channels=3, name="jpeg_reader")
    float_caster = tf.cast(image_reader, tf.float32)
    dims_expander = tf.expand_dims(float_caster, 0)
    resized = tf.image.resize_bilinear(dims_expander, [height, width])
    normalized = tf.divide(tf.subtract(resized, [0]), [255])
    sess = tf.Session()
    result = sess.run(normalized)
    return result

# classify_image("http://gajamed.pl/uploads/2016/04/29.jpg)
# => [("daisy", 0.2), ("rose", 0.9)]
def classify_image(url):
    labels = map(str.strip, open("output_labels.txt").readlines())
    graph = load_graph("output_graph.pb")
    input = graph.get_operation_by_name("import/Placeholder").outputs[0]
    output = graph.get_operation_by_name("import/final_result").outputs[0]
    filepath = download(url)
    image = read_image(filepath)
    with tf.Session(graph=graph) as sess:
        results = sess.run(output, { input: image })[0]
    return list(zip(labels, results))

  
  
api = responder.API()

# curl http://localhost:5432/http://gajamed.pl/uploads/2016/04/29.jpg
#   rose: 90%
#   daisy: 20%
@api.route(None, default=True)
def handler(req, resp):
    image_url = req.url.path[1:]

    text = ""
    for label, confidence in classify_image(image_url):
        text += label + ": " + str(int(confidence * 100)) + "%\n"
    resp.text = text

# You can run this script by first activating anaconda
# and then running python index.py
# Api should be available at the address visible in terminal
if __name__ == '__main__':
    api.run()