video tutorials

README.rst

@@ -406,6 +406,10 @@ Advanced
 .. _pythondgenerator: https://github.com/instillai/TensorFlow-Course/blob/master/codes/python/advanced/dataset_generator.py
 .. _videodgenerator: https://youtu.be/-YsgMdDPu3g
 
+.. _ipythontfrecords: https://github.com/instillai/TensorFlow-Course/blob/master/codes/ipython/advanced/tfrecords.ipynb
+.. _pythontfrecords: https://github.com/instillai/TensorFlow-Course/blob/master/codes/python/advanced/tfrecords.py
+.. _videotfrecords: https://youtu.be/zqavy_5QMk8
+
 
 .. |ctraining| image:: https://colab.research.google.com/assets/colab-badge.svg
  :target: https://colab.research.google.com/github/instillai/TensorFlow-Course/blob/master/codes/ipython/advanced/custom_training.ipynb
@@ -413,6 +417,9 @@ Advanced
 .. |dgenerator| image:: https://colab.research.google.com/assets/colab-badge.svg
   :target: https://colab.research.google.com/github/instillai/TensorFlow-Course/blob/master/codes/ipython/advanced/dataset_generator.ipynb
 
+.. |dgenerator| image:: https://colab.research.google.com/assets/colab-badge.svg
+  :target: https://colab.research.google.com/github/instillai/TensorFlow-Course/blob/master/codes/ipython/advanced/tfrecords.ipynb
+
 
 +----+------------------------------------------+--------------------------+--------------------------------------------------------------------+----------------------------------------+
 | #  |       topic                              |          Run             |  Source Code                                                       |           Media                        |
@@ -421,6 +428,8 @@ Advanced
 +----+------------------------------------------+--------------------------+--------------------------------------------------------------------+----------------------------------------+
 | 2  |  *Dataset Generator*                     |       |dgenerator|       | `Notebook <ipythondgenerator_>`_ / `Python <pythondgenerator_>`_   | `Video Tutorial <videodgenerator_>`_   |
 +----+------------------------------------------+--------------------------+--------------------------------------------------------------------+----------------------------------------+
+| 2  |  *Create TFRecords*                      |       |tfrecords|        | `Notebook <ipythontfrecords_>`_ / `Python <pythontfrecords_>`_     | `Video Tutorial <videotfrecords_>`_   |
++----+------------------------------------------+--------------------------+--------------------------------------------------------------------+----------------------------------------+
 
 
 

codes/python/advanced/tfrecords.py

+# -*- coding: utf-8 -*-
+"""TFRecords.ipynb
+
+Automatically generated by Colaboratory.
+
+Original file is located at
+    https://colab.research.google.com/drive/1p-Nz6v3CyqKSc-QazX1FgvZkamt5T-uC
+"""
+
+import tensorflow as tf
+from tensorflow import keras
+import numpy as np
+
+# Load MNIST data
+(x_train, y_train), (x_test, y_test) = keras.datasets.mnist.load_data()
+
+# Preprocessing
+x_train = x_train / 255.0
+x_test = x_test / 255.0
+
+# Track the data type
+dataType = x_train.dtype
+print(f"Data type: {dataType}")
+
+labelType = y_test.dtype
+print(f"Data type: {labelType}")
+
+im_list = []
+n_samples_to_show = 16
+c = 0
+for i in range(n_samples_to_show):
+  im_list.append(x_train[i])
+
+# Visualization
+import matplotlib.pyplot as plt
+from mpl_toolkits.axes_grid1 import ImageGrid
+fig = plt.figure(figsize=(4., 4.))
+# Ref: https://matplotlib.org/3.1.1/gallery/axes_grid1/simple_axesgrid.html
+grid = ImageGrid(fig, 111,  # similar to subplot(111)
+                 nrows_ncols=(4, 4),  # creates 2x2 grid of axes
+                 axes_pad=0.1,  # pad between axes in inch.
+                 )
+# Show image grid
+for ax, im in zip(grid, im_list):
+    # Iterating over the grid returns the Axes.
+    ax.imshow(im, 'gray')
+plt.show()
+
+# Convert values to compatible tf.Example types.
+
+def _bytes_feature(value):
+    """Returns a bytes_list from a string / byte."""
+    if isinstance(value, type(tf.constant(0))):
+        value = value.numpy()  # BytesList won't unpack a string from an EagerTensor.
+    return tf.train.Feature(bytes_list=tf.train.BytesList(value=[value]))
+
+
+def _float_feature(value):
+    """Returns a float_list from a float / double."""
+    return tf.train.Feature(float_list=tf.train.FloatList(value=[value]))
+
+
+def _int64_feature(value):
+    """Returns an int64_list from a bool / enum / int / uint."""
+    return tf.train.Feature(int64_list=tf.train.Int64List(value=[value]))
+
+# Create the features dictionary.
+def image_example(image, label, dimension):
+    feature = {
+        'dimension': _int64_feature(dimension),
+        'label': _int64_feature(label),
+        'image_raw': _bytes_feature(image.tobytes()),
+    }
+
+    return tf.train.Example(features=tf.train.Features(feature=feature))
+
+record_file = 'mnistTrain.tfrecords'
+n_samples = x_train.shape[0]
+dimension = x_train.shape[1]
+
+with tf.io.TFRecordWriter(record_file) as writer:
+   for i in range(n_samples):
+      image = x_train[i]
+      label = y_train[i]
+      tf_example = image_example(image, label, dimension)
+      writer.write(tf_example.SerializeToString())
+
+# Create the dataset object from tfrecord file(s)
+dataset = tf.data.TFRecordDataset(record_file)
+
+# Decoding function
+def parse_record(record):
+    name_to_features = {
+        'dimension': tf.io.FixedLenFeature([], tf.int64),
+        'label': tf.io.FixedLenFeature([], tf.int64),
+        'image_raw': tf.io.FixedLenFeature([], tf.string),
+    }
+    return tf.io.parse_single_example(record, name_to_features)
+
+def decode_record(record):
+    image = tf.io.decode_raw(
+        record['image_raw'], out_type=dataType, little_endian=True, fixed_length=None, name=None
+    )
+    label = record['label']
+    dimension = record['dimension']
+    image = tf.reshape(image, (dimension, dimension))
+
+    return (image, label)
+
+im_list = []
+n_samples_to_show = 16
+c = 0
+for record in dataset:
+  c+=1
+  if c > n_samples_to_show:
+    break
+  parsed_record = parse_record(record)
+  decoded_record = decode_record(parsed_record)
+  image, label = decoded_record
+  im_list.append(image)
+
+# Visualization
+import matplotlib.pyplot as plt
+from mpl_toolkits.axes_grid1 import ImageGrid
+fig = plt.figure(figsize=(4., 4.))
+# Ref: https://matplotlib.org/3.1.1/gallery/axes_grid1/simple_axesgrid.html
+grid = ImageGrid(fig, 111,  # similar to subplot(111)
+                 nrows_ncols=(4, 4),  # creates 2x2 grid of axes
+                 axes_pad=0.1,  # pad between axes in inch.
+                 )
+# Show image grid
+for ax, im in zip(grid, im_list):
+    # Iterating over the grid returns the Axes.
+    ax.imshow(im, 'gray')
+plt.show()
+