Code upgrade for TensorFlow 2

README.md

@@ -1,4 +1,5 @@
 # Overview
+#"TensorFlow for poets 2" for TensorFlow 2.0
 
 This repo contains code for the "TensorFlow for poets 2" series of codelabs.
 

scripts/count_ops.py

@@ -25,7 +25,7 @@
 def load_graph(file_name):
   with open(file_name,'rb') as f:
     content = f.read()
-  graph_def = tf.GraphDef()
+  graph_def = tf.compat.v1.GraphDef()
   graph_def.ParseFromString(content)
   with tf.Graph().as_default() as graph:
     tf.import_graph_def(graph_def, name='')

scripts/evaluate.py

@@ -31,16 +31,16 @@
 
 def evaluate_graph(graph_file_name):
     with load_graph(graph_file_name).as_default() as graph:
-        ground_truth_input = tf.placeholder(
+        ground_truth_input = tf.compat.v1.placeholder(
             tf.float32, [None, 5], name='GroundTruthInput')
 
         image_buffer_input = graph.get_tensor_by_name('input:0')
         final_tensor = graph.get_tensor_by_name('final_result:0')
         accuracy, _ = retrain.add_evaluation_step(final_tensor, ground_truth_input)
 
         logits = graph.get_tensor_by_name("final_training_ops/Wx_plus_b/add:0")
-        xent = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(
-            labels = ground_truth_input,
+        xent = tf.reduce_mean(input_tensor=tf.nn.softmax_cross_entropy_with_logits(
+            labels = tf.stop_gradient( ground_truth_input),
             logits = logits))
 
     image_dir = 'tf_files/flower_photos'
@@ -68,7 +68,7 @@ def evaluate_graph(graph_file_name):
 
     accuracies = []
     xents = []
-    with tf.Session(graph=graph) as sess:
+    with tf.compat.v1.Session(graph=graph) as sess:
         for filename, ground_truth in zip(filenames, ground_truths):    
             image = Image.open(filename).resize((224,224),Image.ANTIALIAS)
             image = np.array(image, dtype=np.float32)[None,...]

scripts/graph_pb2tb.py

@@ -22,16 +22,16 @@
 def load_graph(graph_pb_path):
   with open(graph_pb_path,'rb') as f:
     content = f.read()
-  graph_def = tf.GraphDef()
+  graph_def = tf.compat.v1.GraphDef()
   graph_def.ParseFromString(content)
   with tf.Graph().as_default() as graph:
     tf.import_graph_def(graph_def, name='')
   return graph
 
 
 def graph_to_tensorboard(graph, out_dir):
-  with tf.Session():
-    train_writer = tf.summary.FileWriter(out_dir)
+  with tf.compat.v1.Session():
+    train_writer = tf.compat.v1.summary.FileWriter(out_dir)
     train_writer.add_graph(graph)
 
 

scripts/label_image.py

@@ -23,10 +23,11 @@
 
 import numpy as np
 import tensorflow as tf
+tf.compat.v1.disable_eager_execution()
 
 def load_graph(model_file):
   graph = tf.Graph()
-  graph_def = tf.GraphDef()
+  graph_def = tf.compat.v1.GraphDef()
 
   with open(model_file, "rb") as f:
     graph_def.ParseFromString(f.read())
@@ -39,7 +40,7 @@ def read_tensor_from_image_file(file_name, input_height=299, input_width=299,
 				input_mean=0, input_std=255):
   input_name = "file_reader"
   output_name = "normalized"
-  file_reader = tf.read_file(file_name, input_name)
+  file_reader = tf.io.read_file(file_name, input_name)
   if file_name.endswith(".png"):
     image_reader = tf.image.decode_png(file_reader, channels = 3,
                                        name='png_reader')
@@ -53,16 +54,16 @@ def read_tensor_from_image_file(file_name, input_height=299, input_width=299,
                                         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, [input_height, input_width])
+  resized = tf.image.resize(dims_expander, [input_height, input_width], method=tf.image.ResizeMethod.BILINEAR)
   normalized = tf.divide(tf.subtract(resized, [input_mean]), [input_std])
-  sess = tf.Session()
+  sess = tf.compat.v1.Session()
   result = sess.run(normalized)
 
   return result
 
 def load_labels(label_file):
   label = []
-  proto_as_ascii_lines = tf.gfile.GFile(label_file).readlines()
+  proto_as_ascii_lines = tf.io.gfile.GFile(label_file).readlines()
   for l in proto_as_ascii_lines:
     label.append(l.rstrip())
   return label
@@ -71,11 +72,11 @@ def load_labels(label_file):
   file_name = "tf_files/flower_photos/daisy/3475870145_685a19116d.jpg"
   model_file = "tf_files/retrained_graph.pb"
   label_file = "tf_files/retrained_labels.txt"
-  input_height = 224
-  input_width = 224
+  input_height = 299
+  input_width = 299
   input_mean = 128
   input_std = 128
-  input_layer = "input"
+  input_layer = "Mul"
   output_layer = "final_result"
 
   parser = argparse.ArgumentParser()
@@ -121,7 +122,7 @@ def load_labels(label_file):
   input_operation = graph.get_operation_by_name(input_name);
   output_operation = graph.get_operation_by_name(output_name);
 
-  with tf.Session(graph=graph) as sess:
+  with tf.compat.v1.Session(graph=graph) as sess:
     start = time.time()
     results = sess.run(output_operation.outputs[0],
                       {input_operation.outputs[0]: t})

scripts/quantize_graph.py

@@ -31,6 +31,7 @@
 import collections
 import re
 import numpy as np
+import tensorflow as tf
 
 from tensorflow.core.framework import attr_value_pb2
 from tensorflow.core.framework import graph_pb2
@@ -584,7 +585,7 @@ def eightbitize_nodes_recursively(self, current_node):
       quantize_input = False
       if current_node.op in ("MatMul", "Conv2D", "BiasAdd", "MaxPool",
                              "AvgPool", "Relu", "Relu6",
-                             "BatchNormWithGlobalNormalization"):
+                             tf.nn.batch_normalization()):
         quantize_input = True
       elif current_node.op == "Concat" and i > 0:
         quantize_input = (
@@ -616,7 +617,7 @@ def eightbitize_nodes_recursively(self, current_node):
     elif (current_node.op == "Concat" and
           dtypes.as_dtype(current_node.attr["T"].type) == dtypes.float32):
       self.eightbitize_concat_node(current_node)
-    elif current_node.op == "BatchNormWithGlobalNormalization":
+    elif current_node.op == tf.nn.batch_normalization():
       self.eightbitize_batch_norm_node(current_node)
     elif (current_node.op == "Reshape" and
           dtypes.as_dtype(current_node.attr["T"].type) == dtypes.float32):
@@ -1044,7 +1045,7 @@ def eightbitize_batch_norm_node(self, original_node):
         self.eightbitize_input_to_node(namespace_prefix, original_gamma_name,
                                        reshape_dims_name, reduction_dims_name))
     quantized_batch_norm_node = create_node(
-        "QuantizedBatchNormWithGlobalNormalization", quantized_batch_norm_name,
+        tf.nn.batch_normalization(), quantized_batch_norm_name,
         [
             quantize_input_name, min_input_name, max_input_name,
             quantize_mean_name, min_mean_name, max_mean_name,
@@ -1187,7 +1188,7 @@ def apply_final_node_renames(self):
   def remove_dead_nodes(self, output_names):
     """Removes nodes that are no longer needed for inference from the graph."""
     old_output_graph = self.output_graph
-    self.output_graph = graph_util.extract_sub_graph(old_output_graph,
+    self.output_graph = tf.compat.v1.graph_util.extract_sub_graph(old_output_graph,
                                                      output_names)
 
   def quantize_weights(self, input_graph, quantization_mode):