modify example with analytics-zoo

tensorflow/image_recognition/classify_image.py

@@ -1,4 +1,5 @@
-# Copyright 2015 The TensorFlow Authors. All Rights Reserved.
+# Copyright 2015 The TensorFlow Authors, 2019 Analytics Zoo Authors. 
+# All Rights Reserved.
 #
 # Licensed under the Apache License, Version 2.0 (the "License");
 # you may not use this file except in compliance with the License.
@@ -44,6 +45,8 @@
 import numpy as np
 from six.moves import urllib
 import tensorflow as tf
+from zoo import init_nncontext
+from zoo.pipeline.api.net import TFDataset, TFPredictor
 
 FLAGS = None
 
@@ -117,14 +120,27 @@ def id_to_string(self, node_id):
     return self.node_lookup[node_id]
 
 
-def create_graph():
+def create_graph(image_array):
   """Creates a graph from saved GraphDef file and returns a saver."""
   # Creates graph from saved graph_def.pb.
+  h, w, c = image_array.shape
+  # get the TFDataset
+  sc = init_nncontext()
+  image_rdd = sc.parallelize(image_array[None, ...]).map(lambda x: [x])
+  image_dataset = TFDataset.from_rdd(image_rdd,
+                                     names=['features'],
+                                     shapes=[[w, h, c]],
+                                     types=[tf.uint8],
+                                     batch_per_thread=1,
+                                     hard_code_batch_size=True)
+  image_tensor = image_dataset.tensors[0]
   with tf.gfile.FastGFile(os.path.join(
       FLAGS.model_dir, 'classify_image_graph_def.pb'), 'rb') as f:
     graph_def = tf.GraphDef()
     graph_def.ParseFromString(f.read())
-    _ = tf.import_graph_def(graph_def, name='')
+    _ = tf.import_graph_def(graph_def,
+                            input_map={'DecodeJpeg:0': image_tensor[0]},
+                            name='')
 
 
 def run_inference_on_image(image):
@@ -138,10 +154,10 @@ def run_inference_on_image(image):
   """
   if not tf.gfile.Exists(image):
     tf.logging.fatal('File does not exist %s', image)
-  image_data = tf.gfile.FastGFile(image, 'rb').read()
-
+  from PIL import Image
+  image_array = np.array(Image.open(image))[:, :, 0:3]
   # Creates graph from saved GraphDef.
-  create_graph()
+  create_graph(image_array)
 
   with tf.Session() as sess:
     # Some useful tensors:
@@ -153,8 +169,9 @@ def run_inference_on_image(image):
     #   encoding of the image.
     # Runs the softmax tensor by feeding the image_data as input to the graph.
     softmax_tensor = sess.graph.get_tensor_by_name('softmax:0')
-    predictions = sess.run(softmax_tensor,
-                           {'DecodeJpeg/contents:0': image_data})
+    predictor = TFPredictor(sess, [softmax_tensor])
+    predictions = predictor.predict().collect()
+
     predictions = np.squeeze(predictions)
 
     # Creates node ID --> English string lookup.