TensorFlow Input Data with Queue

  1. Input raw data (e.g., text, image address). There are many different implementations of queue.

  2. A more elegant way is converting raw data to tfrecord format.

Tips:

  1. setting large number_of_threading (e.g., 10) is helpful.

  2. shuffle the training samples to avoid homogenuity when necessary.

  3. place the training data in local disk instead of removable disk (consider I/O speed).