Python dataset_builder.build方法代码示例

# 需要导入模块: from object_detection.builders import dataset_builder [as 别名]
# 或者: from object_detection.builders.dataset_builder import build [as 别名]
def test_build_tf_record_input_reader_and_load_instance_masks(self):
    tf_record_path = self.create_tf_record()

    input_reader_text_proto = """
      shuffle: false
      num_readers: 1
      load_instance_masks: true
      tf_record_input_reader {{
        input_path: '{0}'
      }}
    """.format(tf_record_path)
    input_reader_proto = input_reader_pb2.InputReader()
    text_format.Merge(input_reader_text_proto, input_reader_proto)
    tensor_dict = dataset_builder.make_initializable_iterator(
        dataset_builder.build(input_reader_proto, batch_size=1)).get_next()

    with tf.train.MonitoredSession() as sess:
      output_dict = sess.run(tensor_dict)
    self.assertAllEqual(
        (1, 1, 4, 5),
        output_dict[fields.InputDataFields.groundtruth_instance_masks].shape) 

# 需要导入模块: from object_detection.builders import dataset_builder [as 别名]
# 或者: from object_detection.builders.dataset_builder import build [as 别名]
def test_sample_all_data(self):
    tf_record_path = self.create_tf_record(num_examples=2)

    input_reader_text_proto = """
      shuffle: false
      num_readers: 1
      sample_1_of_n_examples: 1
      tf_record_input_reader {{
        input_path: '{0}'
      }}
    """.format(tf_record_path)
    input_reader_proto = input_reader_pb2.InputReader()
    text_format.Merge(input_reader_text_proto, input_reader_proto)
    tensor_dict = dataset_builder.make_initializable_iterator(
        dataset_builder.build(input_reader_proto, batch_size=1)).get_next()

    with tf.train.MonitoredSession() as sess:
      output_dict = sess.run(tensor_dict)
      self.assertAllEqual(['0'], output_dict[fields.InputDataFields.source_id])
      output_dict = sess.run(tensor_dict)
      self.assertEquals(['1'], output_dict[fields.InputDataFields.source_id]) 

# 需要导入模块: from object_detection.builders import dataset_builder [as 别名]
# 或者: from object_detection.builders.dataset_builder import build [as 别名]
def test_sample_one_of_n_shards(self):
    tf_record_path = self.create_tf_record(num_examples=4)

    input_reader_text_proto = """
      shuffle: false
      num_readers: 1
      sample_1_of_n_examples: 2
      tf_record_input_reader {{
        input_path: '{0}'
      }}
    """.format(tf_record_path)
    input_reader_proto = input_reader_pb2.InputReader()
    text_format.Merge(input_reader_text_proto, input_reader_proto)
    tensor_dict = dataset_builder.make_initializable_iterator(
        dataset_builder.build(input_reader_proto, batch_size=1)).get_next()

    with tf.train.MonitoredSession() as sess:
      output_dict = sess.run(tensor_dict)
      self.assertAllEqual(['0'], output_dict[fields.InputDataFields.source_id])
      output_dict = sess.run(tensor_dict)
      self.assertEquals(['2'], output_dict[fields.InputDataFields.source_id]) 

# 需要导入模块: from object_detection.builders import dataset_builder [as 别名]
# 或者: from object_detection.builders.dataset_builder import build [as 别名]
def test_build_tf_record_input_reader_and_load_instance_masks(self):
    tf_record_path = self.create_tf_record()

    input_reader_text_proto = """
      shuffle: false
      num_readers: 1
      load_instance_masks: true
      tf_record_input_reader {{
        input_path: '{0}'
      }}
    """.format(tf_record_path)
    input_reader_proto = input_reader_pb2.InputReader()
    text_format.Merge(input_reader_text_proto, input_reader_proto)
    tensor_dict = dataset_util.make_initializable_iterator(
        dataset_builder.build(input_reader_proto, batch_size=1)).get_next()

    sv = tf.train.Supervisor(logdir=self.get_temp_dir())
    with sv.prepare_or_wait_for_session() as sess:
      sv.start_queue_runners(sess)
      output_dict = sess.run(tensor_dict)
    self.assertAllEqual(
        (1, 1, 4, 5),
        output_dict[fields.InputDataFields.groundtruth_instance_masks].shape) 

# 需要导入模块: from object_detection.builders import dataset_builder [as 别名]
# 或者: from object_detection.builders.dataset_builder import build [as 别名]
def test_build_tf_record_input_reader_with_additional_channels(self):
    tf_record_path = self.create_tf_record(has_additional_channels=True)

    input_reader_text_proto = """
      shuffle: false
      num_readers: 1
      tf_record_input_reader {{
        input_path: '{0}'
      }}
    """.format(tf_record_path)
    input_reader_proto = input_reader_pb2.InputReader()
    text_format.Merge(input_reader_text_proto, input_reader_proto)
    tensor_dict = dataset_util.make_initializable_iterator(
        dataset_builder.build(
            input_reader_proto, batch_size=2,
            num_additional_channels=2)).get_next()

    sv = tf.train.Supervisor(logdir=self.get_temp_dir())
    with sv.prepare_or_wait_for_session() as sess:
      sv.start_queue_runners(sess)
      output_dict = sess.run(tensor_dict)

    self.assertEquals((2, 4, 5, 5),
                      output_dict[fields.InputDataFields.image].shape) 

# 需要导入模块: from object_detection.builders import dataset_builder [as 别名]
# 或者: from object_detection.builders.dataset_builder import build [as 别名]
def evaluate(self, eval_pipeline_file, model_dir, eval_dir):
        configs = self._get_configs_from_pipeline_file(eval_pipeline_file)
        model_config = configs['model']
        eval_config = configs['eval_config']
        input_config = configs['eval_input_config']
        model_fn = functools.partial(
            model_builder.build,
            model_config=model_config,
            is_training=True)
        create_input_dict_fn = functools.partial(self.get_next, input_config)
        label_map = label_map_util.load_labelmap(input_config.label_map_path)
        max_num_classes = max([item.id for item in label_map.item])
        categories = label_map_util.convert_label_map_to_categories(
                        label_map, max_num_classes)
        evaluator.evaluate(create_input_dict_fn, model_fn, eval_config, categories,
                        model_dir, eval_dir) 

# 需要导入模块: from object_detection.builders import dataset_builder [as 别名]
# 或者: from object_detection.builders.dataset_builder import build [as 别名]
def test_build_tf_record_input_reader_and_load_instance_masks(self):
    tf_record_path = self.create_tf_record()

    input_reader_text_proto = """
      shuffle: false
      num_readers: 1
      load_instance_masks: true
      tf_record_input_reader {{
        input_path: '{0}'
      }}
    """.format(tf_record_path)
    input_reader_proto = input_reader_pb2.InputReader()
    text_format.Merge(input_reader_text_proto, input_reader_proto)
    tensor_dict = dataset_util.make_initializable_iterator(
        dataset_builder.build(input_reader_proto)).get_next()

    sv = tf.train.Supervisor(logdir=self.get_temp_dir())
    with sv.prepare_or_wait_for_session() as sess:
      sv.start_queue_runners(sess)
      output_dict = sess.run(tensor_dict)
    self.assertAllEqual(
        (1, 4, 5),
        output_dict[fields.InputDataFields.groundtruth_instance_masks].shape) 

# 需要导入模块: from object_detection.builders import dataset_builder [as 别名]
# 或者: from object_detection.builders.dataset_builder import build [as 别名]
def test_build_tf_record_input_reader_and_load_instance_masks(self):
    tf_record_path = self.create_tf_record()

    input_reader_text_proto = """
      shuffle: false
      num_readers: 1
      load_instance_masks: true
      tf_record_input_reader {{
        input_path: '{0}'
      }}
    """.format(tf_record_path)
    input_reader_proto = input_reader_pb2.InputReader()
    text_format.Merge(input_reader_text_proto, input_reader_proto)
    tensor_dict = dataset_builder.make_initializable_iterator(
        dataset_builder.build(input_reader_proto, batch_size=1)).get_next()

    sv = tf.train.Supervisor(logdir=self.get_temp_dir())
    with sv.prepare_or_wait_for_session() as sess:
      sv.start_queue_runners(sess)
      output_dict = sess.run(tensor_dict)
    self.assertAllEqual(
        (1, 1, 4, 5),
        output_dict[fields.InputDataFields.groundtruth_instance_masks].shape) 

# 需要导入模块: from object_detection.builders import dataset_builder [as 别名]
# 或者: from object_detection.builders.dataset_builder import build [as 别名]
def augment_input_data(tensor_dict, data_augmentation_options):
  """Applies data augmentation ops to input tensors.

  Args:
    tensor_dict: A dictionary of input tensors keyed by fields.InputDataFields.
    data_augmentation_options: A list of tuples, where each tuple contains a
      function and a dictionary that contains arguments and their values.
      Usually, this is the output of core/preprocessor.build.

  Returns:
    A dictionary of tensors obtained by applying data augmentation ops to the
    input tensor dictionary.
  """
  tensor_dict[fields.InputDataFields.image] = tf.expand_dims(
      tf.to_float(tensor_dict[fields.InputDataFields.image]), 0)

  include_instance_masks = (fields.InputDataFields.groundtruth_instance_masks
                            in tensor_dict)
  include_keypoints = (fields.InputDataFields.groundtruth_keypoints
                       in tensor_dict)
  tensor_dict = preprocessor.preprocess(
      tensor_dict, data_augmentation_options,
      func_arg_map=preprocessor.get_default_func_arg_map(
          include_instance_masks=include_instance_masks,
          include_keypoints=include_keypoints))
  tensor_dict[fields.InputDataFields.image] = tf.squeeze(
      tensor_dict[fields.InputDataFields.image], axis=0)
  return tensor_dict 

# 需要导入模块: from object_detection.builders import dataset_builder [as 别名]
# 或者: from object_detection.builders.dataset_builder import build [as 别名]
def test_build_tf_record_input_reader(self):
    tf_record_path = self.create_tf_record()

    input_reader_text_proto = """
      shuffle: false
      num_readers: 1
      tf_record_input_reader {{
        input_path: '{0}'
      }}
    """.format(tf_record_path)
    input_reader_proto = input_reader_pb2.InputReader()
    text_format.Merge(input_reader_text_proto, input_reader_proto)
    tensor_dict = dataset_builder.make_initializable_iterator(
        dataset_builder.build(input_reader_proto, batch_size=1)).get_next()

    with tf.train.MonitoredSession() as sess:
      output_dict = sess.run(tensor_dict)

    self.assertTrue(
        fields.InputDataFields.groundtruth_instance_masks not in output_dict)
    self.assertEquals((1, 4, 5, 3),
                      output_dict[fields.InputDataFields.image].shape)
    self.assertAllEqual([[2]],
                        output_dict[fields.InputDataFields.groundtruth_classes])
    self.assertEquals(
        (1, 1, 4), output_dict[fields.InputDataFields.groundtruth_boxes].shape)
    self.assertAllEqual(
        [0.0, 0.0, 1.0, 1.0],
        output_dict[fields.InputDataFields.groundtruth_boxes][0][0]) 

# 需要导入模块: from object_detection.builders import dataset_builder [as 别名]
# 或者: from object_detection.builders.dataset_builder import build [as 别名]
def test_build_tf_record_input_reader_with_batch_size_two_and_masks(self):
    tf_record_path = self.create_tf_record()

    input_reader_text_proto = """
      shuffle: false
      num_readers: 1
      load_instance_masks: true
      tf_record_input_reader {{
        input_path: '{0}'
      }}
    """.format(tf_record_path)
    input_reader_proto = input_reader_pb2.InputReader()
    text_format.Merge(input_reader_text_proto, input_reader_proto)

    def one_hot_class_encoding_fn(tensor_dict):
      tensor_dict[fields.InputDataFields.groundtruth_classes] = tf.one_hot(
          tensor_dict[fields.InputDataFields.groundtruth_classes] - 1, depth=3)
      return tensor_dict

    tensor_dict = dataset_builder.make_initializable_iterator(
        dataset_builder.build(
            input_reader_proto,
            transform_input_data_fn=one_hot_class_encoding_fn,
            batch_size=2)).get_next()

    with tf.train.MonitoredSession() as sess:
      output_dict = sess.run(tensor_dict)

    self.assertAllEqual(
        [2, 1, 4, 5],
        output_dict[fields.InputDataFields.groundtruth_instance_masks].shape) 

# 需要导入模块: from object_detection.builders import dataset_builder [as 别名]
# 或者: from object_detection.builders.dataset_builder import build [as 别名]
def test_raises_error_with_no_input_paths(self):
    input_reader_text_proto = """
      shuffle: false
      num_readers: 1
      load_instance_masks: true
    """
    input_reader_proto = input_reader_pb2.InputReader()
    text_format.Merge(input_reader_text_proto, input_reader_proto)
    with self.assertRaises(ValueError):
      dataset_builder.build(input_reader_proto, batch_size=1) 

# 需要导入模块: from object_detection.builders import dataset_builder [as 别名]
# 或者: from object_detection.builders.dataset_builder import build [as 别名]
def test_build_tf_record_input_reader(self):
    tf_record_path = self.create_tf_record()

    input_reader_text_proto = """
      shuffle: false
      num_readers: 1
      tf_record_input_reader {{
        input_path: '{0}'
      }}
    """.format(tf_record_path)
    input_reader_proto = input_reader_pb2.InputReader()
    text_format.Merge(input_reader_text_proto, input_reader_proto)
    tensor_dict = dataset_util.make_initializable_iterator(
        dataset_builder.build(input_reader_proto, batch_size=1)).get_next()

    sv = tf.train.Supervisor(logdir=self.get_temp_dir())
    with sv.prepare_or_wait_for_session() as sess:
      sv.start_queue_runners(sess)
      output_dict = sess.run(tensor_dict)

    self.assertTrue(
        fields.InputDataFields.groundtruth_instance_masks not in output_dict)
    self.assertEquals((1, 4, 5, 3),
                      output_dict[fields.InputDataFields.image].shape)
    self.assertAllEqual([[2]],
                        output_dict[fields.InputDataFields.groundtruth_classes])
    self.assertEquals(
        (1, 1, 4), output_dict[fields.InputDataFields.groundtruth_boxes].shape)
    self.assertAllEqual(
        [0.0, 0.0, 1.0, 1.0],
        output_dict[fields.InputDataFields.groundtruth_boxes][0][0]) 

# 需要导入模块: from object_detection.builders import dataset_builder [as 别名]
# 或者: from object_detection.builders.dataset_builder import build [as 别名]
def test_build_tf_record_input_reader_with_batch_size_two_and_masks(self):
    tf_record_path = self.create_tf_record()

    input_reader_text_proto = """
      shuffle: false
      num_readers: 1
      load_instance_masks: true
      tf_record_input_reader {{
        input_path: '{0}'
      }}
    """.format(tf_record_path)
    input_reader_proto = input_reader_pb2.InputReader()
    text_format.Merge(input_reader_text_proto, input_reader_proto)

    def one_hot_class_encoding_fn(tensor_dict):
      tensor_dict[fields.InputDataFields.groundtruth_classes] = tf.one_hot(
          tensor_dict[fields.InputDataFields.groundtruth_classes] - 1, depth=3)
      return tensor_dict

    tensor_dict = dataset_util.make_initializable_iterator(
        dataset_builder.build(
            input_reader_proto,
            transform_input_data_fn=one_hot_class_encoding_fn,
            batch_size=2,
            max_num_boxes=2,
            num_classes=3,
            spatial_image_shape=[4, 5])).get_next()

    sv = tf.train.Supervisor(logdir=self.get_temp_dir())
    with sv.prepare_or_wait_for_session() as sess:
      sv.start_queue_runners(sess)
      output_dict = sess.run(tensor_dict)

    self.assertAllEqual(
        [2, 2, 4, 5],
        output_dict[fields.InputDataFields.groundtruth_instance_masks].shape) 

# 需要导入模块: from object_detection.builders import dataset_builder [as 别名]
# 或者: from object_detection.builders.dataset_builder import build [as 别名]
def test_raises_error_with_no_input_paths(self):
    input_reader_text_proto = """
      shuffle: false
      num_readers: 1
      load_instance_masks: true
    """
    input_reader_proto = input_reader_pb2.InputReader()
    text_format.Merge(input_reader_text_proto, input_reader_proto)
    with self.assertRaises(ValueError):
      dataset_builder.build(input_reader_proto)