『tensorflow』ssd源码学习_其八：网络训练

        # =================================================================== #
        # Configure the moving averages.
        # =================================================================== #
        if FLAGS.moving_average_decay:
            moving_average_variables = slim.get_model_variables()
            variable_averages = tf.train.ExponentialMovingAverage(
                FLAGS.moving_average_decay, global_step)
        else:
            moving_average_variables, variable_averages = None, None

        with tf.device(deploy_config.optimizer_device()):
            learning_rate = tf_utils.configure_learning_rate(FLAGS,
                                                             dataset.num_samples,
                                                             global_step)

def configure_learning_rate(flags, num_samples_per_epoch, global_step):
    """Configures the learning rate.

    Args:
      num_samples_per_epoch: The number of samples in each epoch of training.
      global_step: The global_step tensor.
    Returns:
      A `Tensor` representing the learning rate.
    """
    decay_steps = int(num_samples_per_epoch / flags.batch_size *
                      flags.num_epochs_per_decay)

    if flags.learning_rate_decay_type == ‘exponential‘:
        return tf.train.exponential_decay(flags.learning_rate,
                                          global_step,
                                          decay_steps,
                                          flags.learning_rate_decay_factor,
                                          staircase=True,
                                          name=‘exponential_decay_learning_rate‘)
    elif flags.learning_rate_decay_type == ‘fixed‘:
        return tf.constant(flags.learning_rate, name=‘fixed_learning_rate‘)
    elif flags.learning_rate_decay_type == ‘polynomial‘:
        return tf.train.polynomial_decay(flags.learning_rate,
                                         global_step,
                                         decay_steps,
                                         flags.end_learning_rate,
                                         power=1.0,
                                         cycle=False,
                                         name=‘polynomial_decay_learning_rate‘)

optimizer = tf_utils.configure_optimizer(FLAGS, learning_rate)

def configure_optimizer(flags, learning_rate):
    """Configures the optimizer used for training.

    Args:
      learning_rate: A scalar or `Tensor` learning rate.
    Returns:
      An instance of an optimizer.
    """
    if flags.optimizer == ‘adadelta‘:
        optimizer = tf.train.AdadeltaOptimizer(
            learning_rate,
            rho=flags.adadelta_rho,
            epsilon=flags.opt_epsilon)
    elif flags.optimizer == ‘adagrad‘:
        optimizer = tf.train.AdagradOptimizer(
            learning_rate,
            initial_accumulator_value=flags.adagrad_initial_accumulator_value)
    elif flags.optimizer == ‘adam‘:
        optimizer = tf.train.AdamOptimizer(
            learning_rate,
            beta1=flags.adam_beta1,
            beta2=flags.adam_beta2,
            epsilon=flags.opt_epsilon)
    elif flags.optimizer == ‘ftrl‘:
        optimizer = tf.train.FtrlOptimizer(
            learning_rate,
            learning_rate_power=flags.ftrl_learning_rate_power,
            initial_accumulator_value=flags.ftrl_initial_accumulator_value,
            l1_regularization_strength=flags.ftrl_l1,
            l2_regularization_strength=flags.ftrl_l2)
    elif flags.optimizer == ‘momentum‘:
        optimizer = tf.train.MomentumOptimizer(
            learning_rate,
            momentum=flags.momentum,
            name=‘Momentum‘)
    elif flags.optimizer == ‘rmsprop‘:
        optimizer = tf.train.RMSPropOptimizer(
            learning_rate,
            decay=flags.rmsprop_decay,
            momentum=flags.rmsprop_momentum,
            epsilon=flags.opt_epsilon)
    elif flags.optimizer == ‘sgd‘:
        optimizer = tf.train.GradientDescentOptimizer(learning_rate)
    else:
        raise ValueError(‘Optimizer [%s] was not recognized‘, flags.optimizer)
    return optimizer

gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=FLAGS.gpu_memory_fraction)
config = tf.ConfigProto(log_device_placement=False,
                        gpu_options=gpu_options)
saver = tf.train.Saver(max_to_keep=5,
                       keep_checkpoint_every_n_hours=1.0,
                       write_version=2,
                       pad_step_number=False)
slim.learning.train(
    train_tensor,
    logdir=FLAGS.train_dir,
    master=‘‘,
    is_chief=True,
    init_fn=tf_utils.get_init_fn(FLAGS),            # 看函数实现就明白了，assign变量用
    summary_op=summary_op,                          # tf.summary.merge节点
    number_of_steps=FLAGS.max_number_of_steps,      # 训练step
    log_every_n_steps=FLAGS.log_every_n_steps,      # 每次model保存step间隔
    save_summaries_secs=FLAGS.save_summaries_secs,  # 每次summary时间间隔
    saver=saver,                                    # tf.train.Saver节点
    save_interval_secs=FLAGS.save_interval_secs,
    session_config=config,                          # sess参数
    sync_optimizer=None)

def get_init_fn(flags):
    """Returns a function run by the chief worker to warm-start the training.
    Note that the init_fn is only run when initializing the model during the very
    first global step.

    Returns:
      An init function run by the supervisor.
    """
    if flags.checkpoint_path is None:
        return None
    # Warn the user if a checkpoint exists in the train_dir. Then ignore.
    if tf.train.latest_checkpoint(flags.train_dir):
        tf.logging.info(
            ‘Ignoring --checkpoint_path because a checkpoint already exists in %s‘
            % flags.train_dir)
        return None

    exclusions = []
    if flags.checkpoint_exclude_scopes:
        exclusions = [scope.strip()
                      for scope in flags.checkpoint_exclude_scopes.split(‘,‘)]

    # TODO(sguada) variables.filter_variables()
    variables_to_restore = []
    for var in slim.get_model_variables():
        excluded = False
        for exclusion in exclusions:
            if var.op.name.startswith(exclusion):
                excluded = True
                break
        if not excluded:
            variables_to_restore.append(var)
    # Change model scope if necessary.
    if flags.checkpoint_model_scope is not None:
        variables_to_restore =             var.op.name.replace(flags.model_name,
                                 flags.checkpoint_model_scope): var
             for var in variables_to_restore


    if tf.gfile.IsDirectory(flags.checkpoint_path):
        checkpoint_path = tf.train.latest_checkpoint(flags.checkpoint_path)
    else:
        checkpoint_path = flags.checkpoint_path
    tf.logging.info(‘Fine-tuning from %s. Ignoring missing vars: %s‘ % (checkpoint_path, flags.ignore_missing_vars))

    return slim.assign_from_checkpoint_fn(
        checkpoint_path,
        variables_to_restore,
        ignore_missing_vars=flags.ignore_missing_vars)