Creating TFRecords Dataset

dataset_Iterator.py

import tensorflow as tf


SHUFFLE_BUFFER = 1000
BATCH_SIZE = 32
NUM_CLASSES = 12

# Create a description of the features.  
feature_description = {
    'feature0': tf.FixedLenFeature([32768], tf.float32),
    'feature1': tf.FixedLenFeature([1], tf.int64)
}

def _parse_function(example_proto):
  # Parse the input tf.Example proto using the dictionary above.
    parsed_example = tf.parse_single_example(example_proto, feature_description)
    parsed_example["feature0"] = tf.transpose(tf.reshape(parsed_example['feature0'], (256,128)))
    return parsed_example

def create_dataset(filepath):
    
    dataset = tf.data.TFRecordDataset(filepath)
    
    dataset = dataset.map(_parse_function) #, num_parallel_calls=8)
    
    # This dataset will go on forever
    dataset = dataset.repeat()
    
    # Set the number of datapoints you want to load and shuffle 
    dataset = dataset.shuffle(SHUFFLE_BUFFER)
    dataset = dataset.batch(BATCH_SIZE)
    
    # Create an iterator
    iterator = dataset.make_one_shot_iterator()
    
    # Create your tf representation of the iterator
    feature = iterator.get_next()
    #print(feature)
    lmfcc = feature["feature0"]
    label = feature["feature1"]
    
    # Bring your picture back in shape
    lmfcc = tf.reshape(lmfcc, [1,128, 256])

    # Create a one hot array for your labels
    label = tf.one_hot(label, NUM_CLASSES)
    label = tf.reshape(label, [1,1, 12])

    print(lmfcc.shape)
    print(label.shape)

    return lmfcc, label