mnist_extension.py

# Copyright 2015 The TensorFlow 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.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ==============================================================================

"""Simple, end-to-end, LeNet-5-like convolutional MNIST model example.

This should achieve a test error of 0.7%. Please keep this model as simple and
linear as possible, it is meant as a tutorial for simple convolutional models.
Run with --self_test on the command line to execute a short self-test.
"""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function

import argparse
import gzip
import numpy as np
import os
import shutil
import sys
import time

import numpy
from six.moves import urllib
from six.moves import xrange  # pylint: disable=redefined-builtin
import tensorflow as tf
from PIL import Image
import itertools
from random import shuffle

FLAGS = None

SOURCE_URL = 'http://yann.lecun.com/exdb/mnist/'
WORK_DIRECTORY = None
IMAGE_SIZE = 28
NUM_CHANNELS = 1
PIXEL_DEPTH = 255
NUM_LABELS = 10
VALIDATION_SIZE = 5000  # Size of the validation set.
SEED = 66478  # Set to None for random seed.
BATCH_SIZE = 64
NUM_EPOCHS = 10
EVAL_BATCH_SIZE = 64
EVAL_FREQUENCY = 100  # Number of steps between evaluations.

EXTENSION_DIR = None

def load_extension_symbol_labels_and_data():
  """Require EXTENSION_DIR exists, and has 6 sub-directories with special names.
     Returns tuple (label: int, data: numpy.array)"""
  assert os.path.isdir(EXTENSION_DIR)

  for it in enumerate(["add", "minus", "mul", "div", "lp", "rp"]):
    label = 10 + it[0]
    sub_dir_name = it[1]

    sub_dir = os.path.join(EXTENSION_DIR, sub_dir_name)
    image_names = os.listdir(sub_dir)
    size = len(image_names)
    print('Loading', size, 'images in "' + sub_dir_name + '"folder.')

    def load_label_data_pair(image_name):
      image_path = os.path.join(sub_dir, image_name)
      img = Image.open(image_path)
      img.load()
      image_data = numpy.asarray(img, dtype=numpy.float32)

      # Convert (Foreground: 0, Background: 255) to (Foreground: 0.5, Background: -0.5), aligned with MNIST
      data = (PIXEL_DEPTH / 2.0 - image_data) / PIXEL_DEPTH
      return (label, data)

    yield from map(load_label_data_pair, image_names)
    #yield from itertools.islice(map(load_label_data_pair, image_names), 0, 50)


def data_type():
  """Return the type of the activations, weights, and placeholder variables."""
  if FLAGS.use_fp16:
    return tf.float16
  else:
    return tf.float32


def maybe_download(filename):
  """Download the data from Yann's website, unless it's already here."""
  if not tf.gfile.Exists(WORK_DIRECTORY):
    tf.gfile.MakeDirs(WORK_DIRECTORY)
  filepath = os.path.join(WORK_DIRECTORY, filename)
  if not tf.gfile.Exists(filepath):
    filepath, _ = urllib.request.urlretrieve(SOURCE_URL + filename, filepath)
    with tf.gfile.GFile(filepath) as f:
      size = f.size()
    print('Successfully downloaded', filename, size, 'bytes.')
  return filepath


def extract_data(filename, num_images):
  """Extract the images into a 4D tensor [image index, y, x, channels].

  Values are rescaled from [0, 255] down to [-0.5, 0.5].
  """
  print('Extracting', filename)
  with gzip.open(filename) as bytestream:
    bytestream.read(16)
    buf = bytestream.read(IMAGE_SIZE * IMAGE_SIZE * num_images * NUM_CHANNELS)
    data = numpy.frombuffer(buf, dtype=numpy.uint8).astype(numpy.float32)
    data = (data - (PIXEL_DEPTH / 2.0)) / PIXEL_DEPTH
    data = data.reshape(num_images, IMAGE_SIZE, IMAGE_SIZE, NUM_CHANNELS)
    return data


def extract_labels(filename, num_images):
  """Extract the labels into a vector of int64 label IDs."""
  print('Extracting', filename)
  with gzip.open(filename) as bytestream:
    bytestream.read(8)
    buf = bytestream.read(1 * num_images)
    labels = numpy.frombuffer(buf, dtype=numpy.uint8).astype(numpy.int64)
  return labels


def fake_data(num_images):
  """Generate a fake dataset that matches the dimensions of MNIST."""
  data = numpy.ndarray(
      shape=(num_images, IMAGE_SIZE, IMAGE_SIZE, NUM_CHANNELS),
      dtype=numpy.float32)
  labels = numpy.zeros(shape=(num_images,), dtype=numpy.int64)
  for image in xrange(num_images):
    label = image % 2
    data[image, :, :, 0] = label - 0.5
    labels[image] = label
  return data, labels


def error_rate(predictions, labels):
  """Return the error rate based on dense predictions and sparse labels."""
  return 100.0 - (
      100.0 *
      numpy.sum(numpy.argmax(predictions, 1) == labels) /
      predictions.shape[0])


def main(_):
  global BATCH_SIZE
  if FLAGS.self_test:
    print('Running self-test.')
    train_data, train_labels = fake_data(256)
    validation_data, validation_labels = fake_data(EVAL_BATCH_SIZE)
    test_data, test_labels = fake_data(EVAL_BATCH_SIZE)
    num_epochs = 1
  else:
    # Get the data.
    train_data_filename = maybe_download('train-images-idx3-ubyte.gz')
    train_labels_filename = maybe_download('train-labels-idx1-ubyte.gz')
    test_data_filename = maybe_download('t10k-images-idx3-ubyte.gz')
    test_labels_filename = maybe_download('t10k-labels-idx1-ubyte.gz')

    # Extract it into numpy arrays.
    train_data = extract_data(train_data_filename, 60000)
    train_labels = extract_labels(train_labels_filename, 60000)
    test_data = extract_data(test_data_filename, 10000)
    test_labels = extract_labels(test_labels_filename, 10000)

    # This is 60000.
    raw_train_size = len(train_data)

    # Load all extension symbols and mix with raw MNIST data
    if EXTENSION_DIR is not None:

      # Including math symbols.
      NUM_LABELS = 16

      extension_label_data_pairs = list(load_extension_symbol_labels_and_data())
      shuffle(extension_label_data_pairs)

      extension_size = len(extension_label_data_pairs)
      test_count = int(extension_size * 0.15) # 15% for test
      train_count = extension_size - test_count

      test_pairs = extension_label_data_pairs[:test_count]
      train_pairs = extension_label_data_pairs[test_count:]

      def mix(pairs, raw_data, raw_labels):
        labels = numpy.asarray([pair[0] for pair in pairs], dtype=numpy.int64)
        data = numpy.asarray([pair[1] for pair in pairs]).reshape((len(pairs), 28, 28, 1))
        labels = numpy.concatenate([raw_labels, labels])
        data = numpy.concatenate([raw_data, data])

        idx = numpy.random.permutation(len(labels))

        return (labels[idx], data[idx])

      train_labels, train_data = mix(train_pairs, train_data, train_labels)
      test_labels, test_data = mix(test_pairs, test_data, test_labels)

      global VALIDATION_SIZE
      validation_ratio = VALIDATION_SIZE * 1.0 / raw_train_size
      VALIDATION_SIZE = VALIDATION_SIZE + int(validation_ratio * train_count)



    # Generate a validation set.
    validation_data = train_data[:VALIDATION_SIZE, ...]
    validation_labels = train_labels[:VALIDATION_SIZE]
    train_data = train_data[VALIDATION_SIZE:, ...]
    train_labels = train_labels[VALIDATION_SIZE:]
    num_epochs = NUM_EPOCHS


  train_size = train_labels.shape[0]

  # This is where training samples and labels are fed to the graph.
  # These placeholder nodes will be fed a batch of training data at each
  # training step using the {feed_dict} argument to the Run() call below.
  train_data_node = tf.placeholder(
      data_type(),
      shape=(None, IMAGE_SIZE, IMAGE_SIZE, NUM_CHANNELS),
      name='image_input')
  train_labels_node = tf.placeholder(tf.int64, shape=(None,))
  eval_data = tf.placeholder(
      data_type(),
      shape=(None, IMAGE_SIZE, IMAGE_SIZE, NUM_CHANNELS))

  # The variables below hold all the trainable weights. They are passed an
  # initial value which will be assigned when we call:
  # {tf.global_variables_initializer().run()}
  conv1_weights = tf.Variable(
      tf.truncated_normal([5, 5, NUM_CHANNELS, 32],  # 5x5 filter, depth 32.
                          stddev=0.1,
                          seed=SEED, dtype=data_type()))
  conv1_biases = tf.Variable(tf.zeros([32], dtype=data_type()))
  conv2_weights = tf.Variable(tf.truncated_normal(
      [5, 5, 32, 64], stddev=0.1,
      seed=SEED, dtype=data_type()))
  conv2_biases = tf.Variable(tf.constant(0.1, shape=[64], dtype=data_type()))
  fc1_weights = tf.Variable(  # fully connected, depth 512.
      tf.truncated_normal([IMAGE_SIZE // 4 * IMAGE_SIZE // 4 * 64, 512],
                          stddev=0.1,
                          seed=SEED,
                          dtype=data_type()))
  fc1_biases = tf.Variable(tf.constant(0.1, shape=[512], dtype=data_type()))
  fc2_weights = tf.Variable(tf.truncated_normal([512, NUM_LABELS],
                                                stddev=0.1,
                                                seed=SEED,
                                                dtype=data_type()))
  fc2_biases = tf.Variable(tf.constant(
      0.1, shape=[NUM_LABELS], dtype=data_type()))

  # We will replicate the model structure for the training subgraph, as well
  # as the evaluation subgraphs, while sharing the trainable parameters.
  def model(data, train=False):
    """The Model definition."""
    # 2D convolution, with 'SAME' padding (i.e. the output feature map has
    # the same size as the input). Note that {strides} is a 4D array whose
    # shape matches the data layout: [image index, y, x, depth].
    conv = tf.nn.conv2d(data,
                        conv1_weights,
                        strides=[1, 1, 1, 1],
                        padding='SAME')
    # Bias and rectified linear non-linearity.
    relu = tf.nn.relu(tf.nn.bias_add(conv, conv1_biases))
    # Max pooling. The kernel size spec {ksize} also follows the layout of
    # the data. Here we have a pooling window of 2, and a stride of 2.
    pool = tf.nn.max_pool(relu,
                          ksize=[1, 2, 2, 1],
                          strides=[1, 2, 2, 1],
                          padding='SAME')
    conv = tf.nn.conv2d(pool,
                        conv2_weights,
                        strides=[1, 1, 1, 1],
                        padding='SAME')
    relu = tf.nn.relu(tf.nn.bias_add(conv, conv2_biases))
    pool = tf.nn.max_pool(relu,
                          ksize=[1, 2, 2, 1],
                          strides=[1, 2, 2, 1],
                          padding='SAME')
    # Reshape the feature map cuboid into a 2D matrix to feed it to the
    # fully connected layers.
    pool_shape = pool.get_shape().as_list()
    reshape = tf.reshape(
        pool,
        [tf.shape(pool)[0], pool_shape[1] * pool_shape[2] * pool_shape[3]])
    # Fully connected layer. Note that the '+' operation automatically
    # broadcasts the biases.
    hidden = tf.nn.relu(tf.matmul(reshape, fc1_weights) + fc1_biases)
    # Add a 50% dropout during training only. Dropout also scales
    # activations such that no rescaling is needed at evaluation time.
    if train:
      hidden = tf.nn.dropout(hidden, 0.5, seed=SEED)
    return tf.matmul(hidden, fc2_weights) + fc2_biases

  # Training computation: logits + cross-entropy loss.
  logits = model(train_data_node, True)
  loss = tf.reduce_mean(tf.nn.sparse_softmax_cross_entropy_with_logits(
      labels=train_labels_node, logits=logits))

  predict_logits = model(train_data_node)
  predict_op = tf.argmax(predict_logits, 1, name='predict_op')

  # L2 regularization for the fully connected parameters.
  regularizers = (tf.nn.l2_loss(fc1_weights) + tf.nn.l2_loss(fc1_biases) +
                  tf.nn.l2_loss(fc2_weights) + tf.nn.l2_loss(fc2_biases))
  # Add the regularization term to the loss.
  loss += 5e-4 * regularizers

  # Optimizer: set up a variable that's incremented once per batch and
  # controls the learning rate decay.
  batch = tf.Variable(0, dtype=data_type())
  # Decay once per epoch, using an exponential schedule starting at 0.01.
  learning_rate = tf.train.exponential_decay(
      0.01,                # Base learning rate.
      batch * BATCH_SIZE,  # Current index into the dataset.
      train_size,          # Decay step.
      0.95,                # Decay rate.
      staircase=True)
  # Use simple momentum for the optimization.
  optimizer = tf.train.MomentumOptimizer(learning_rate,
                                         0.9).minimize(loss,
                                                       global_step=batch)

  # Predictions for the current training minibatch.
  train_prediction = tf.nn.softmax(logits)

  # Predictions for the test and validation, which we'll compute less often.
  eval_prediction = tf.nn.softmax(model(eval_data))

  # Small utility function to evaluate a dataset by feeding batches of data to
  # {eval_data} and pulling the results from {eval_predictions}.
  # Saves memory and enables this to run on smaller GPUs.
  def eval_in_batches(data, sess):
    """Get all predictions for a dataset by running it in small batches."""
    size = data.shape[0]
    if size < EVAL_BATCH_SIZE:
      raise ValueError("batch size for evals larger than dataset: %d" % size)
    predictions = numpy.ndarray(shape=(size, NUM_LABELS), dtype=numpy.float32)
    for begin in xrange(0, size, EVAL_BATCH_SIZE):
      end = begin + EVAL_BATCH_SIZE
      if end <= size:
        predictions[begin:end, :] = sess.run(
            eval_prediction,
            feed_dict={eval_data: data[begin:end, ...]})
      else:
        batch_predictions = sess.run(
            eval_prediction,
            feed_dict={eval_data: data[-EVAL_BATCH_SIZE:, ...]})
        predictions[begin:, :] = batch_predictions[begin - size:, :]
    return predictions

  ### Change original code
  # Add model_dir to save model
  if not os.path.exists(FLAGS.model_dir):
    os.makedirs(FLAGS.model_dir)

  ### Change original code
  # Create a saver for writing training checkpoints.
  saver = tf.train.Saver()
  # Create a builder for writing saved model for serving.
  if os.path.isdir(FLAGS.export_dir):
    shutil.rmtree(FLAGS.export_dir)
  builder = tf.saved_model.builder.SavedModelBuilder(FLAGS.export_dir)


  # Create a local session to run the training.
  start_time = time.time()
  with tf.Session(config=tf.ConfigProto(log_device_placement=False)) as sess:
    # Run all the initializers to prepare the trainable parameters.
    tf.global_variables_initializer().run()
    ### Change original code
    # Save checkpoint when training
    ckpt = tf.train.get_checkpoint_state(FLAGS.model_dir)
    if ckpt and ckpt.model_checkpoint_path:
      print('Load from ' + ckpt.model_checkpoint_path)
      saver.restore(sess, ckpt.model_checkpoint_path)

    ### Change original code
    # Create summary, logs will be saved, which can display in Tensorboard
    tf.summary.scalar("loss", loss)
    merged = tf.summary.merge_all()
    writer = tf.summary.FileWriter(os.path.join(FLAGS.model_dir, 'log'), sess.graph)

    print('Initialized!')

    # Loop through training steps.
    for step in xrange(int(num_epochs * train_size) // BATCH_SIZE):
      # Compute the offset of the current minibatch in the data.
      # Note that we could use better randomization across epochs.
      offset = (step * BATCH_SIZE) % (train_size - BATCH_SIZE)
      batch_data = train_data[offset:(offset + BATCH_SIZE), ...]
      batch_labels = train_labels[offset:(offset + BATCH_SIZE)]
      # This dictionary maps the batch data (as a numpy array) to the
      # node in the graph it should be fed to.
      feed_dict = {train_data_node: batch_data,
                   train_labels_node: batch_labels}
      # Run the optimizer to update weights.
      sess.run(optimizer, feed_dict=feed_dict)
      # print some extra information once reach the evaluation frequency
      if step % EVAL_FREQUENCY == 0:
        # fetch some extra nodes' data
        ### Change original code
        # Add summary
        summary, l, lr, predictions = sess.run([merged, loss, learning_rate, train_prediction],
                                      feed_dict=feed_dict)
        writer.add_summary(summary, step)
        elapsed_time = time.time() - start_time
        start_time = time.time()
        ### Change original code
        # save model
        if step % (EVAL_FREQUENCY * 10) == 0:
          saver.save(sess, os.path.join(FLAGS.model_dir, "model.ckpt"), global_step=step)
        print('Step %d (epoch %.2f), %.1f ms' %
              (step, float(step) * BATCH_SIZE / train_size,
               1000 * elapsed_time / EVAL_FREQUENCY))
        print('Minibatch loss: %.3f, learning rate: %.6f' % (l, lr))
        print('Minibatch error: %.1f%%' % error_rate(predictions, batch_labels))
        print('Validation error: %.1f%%' % error_rate(
            eval_in_batches(validation_data, sess), validation_labels))
        sys.stdout.flush()

    ### Change original code
    # Save model
    inputs = { tf.saved_model.signature_constants.PREDICT_INPUTS: train_data_node }
    outputs = { tf.saved_model.signature_constants.PREDICT_OUTPUTS: predict_op }
    serving_signatures = {
      'Infer': #tf.saved_model.signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY:
      tf.saved_model.signature_def_utils.predict_signature_def(inputs, outputs)
    }
    builder.add_meta_graph_and_variables(sess, [tf.saved_model.tag_constants.SERVING],
                                         signature_def_map=serving_signatures,
                                         assets_collection=tf.get_collection(tf.GraphKeys.ASSET_FILEPATHS),
                                         clear_devices=True)
    builder.save()

    # Finally print the result!
    test_error = error_rate(eval_in_batches(test_data, sess), test_labels)
    print('Test error: %.1f%%' % test_error)
    if FLAGS.self_test:
      print('test_error', test_error)
      assert test_error == 0.0, 'expected 0.0 test_error, got %.2f' % (test_error,)


if __name__ == '__main__':
  parser = argparse.ArgumentParser()
  parser.add_argument(
      '--use_fp16',
      default=False,
      help='Use half floats instead of full floats if True.',
      action='store_true')
  parser.add_argument(
      '--self_test',
      default=False,
      action='store_true',
      help='True if running a self test.')
  parser.add_argument(
    '--input_dir',
    type=str,
    default='input',
    help='Directory to put the input data.')
  parser.add_argument(
    '--model_dir',
    type=str,
    default='output',
    help='Directory to put the checkpoint files.')
  parser.add_argument(
    '--export_dir',
    type=str,
    default='export',
    help='Directory to put the savedmodel files.')

  parser.add_argument(
    '--extension_dir',
    type=str,
    default=None,
    help='Directory to put the extension images.')

  FLAGS, unparsed = parser.parse_known_args()
  WORK_DIRECTORY = FLAGS.input_dir

  EXTENSION_DIR = FLAGS.extension_dir

  tf.app.run(main=main, argv=[sys.argv[0]] + unparsed)