from __future__ import absolute_import, division, print_function, unicode_literals

import tensorflow as tf
# You'll generate plots of attention in order to see which parts of an image
# our model focuses on during captioning
import matplotlib.pyplot as plt
# Scikit-learn includes many helpful utilities
from sklearn.model_selection import train_test_splitfrom sklearn.utils import shuffle
import reimport numpy as npimport osimport timeimport jsonfrom glob import globfrom PIL import Image
import pickle

annotation_zip = tf.keras.utils.get_file('captions.zip',
                                          cache_subdir=os.path.abspath('.'),
                                          origin = 'http://images.cocodataset.org/annotations/annotations_trainval2014.zip',
                                          extract = True)
annotation_file = os.path.dirname(annotation_zip)+'/annotations/captions_train2014.json'

name_of_zip = 'train2014.zip'
if not os.path.exists(os.path.abspath('.') + '/' + name_of_zip):
  image_zip = tf.keras.utils.get_file(name_of_zip,
                                      cache_subdir=os.path.abspath('.'),
                                      origin = 'http://images.cocodataset.org/zips/train2014.zip',
                                      extract = True)
  PATH = os.path.dirname(image_zip)+'/train2014/'
else:
  PATH = os.path.abspath('.')+'/train2014/'

Downloading data from http://images.cocodataset.org/annotations/annotations_trainval2014.zip
252878848/252872794 [==============================] - 8s 0us/step
Downloading data from http://images.cocodataset.org/zips/train2014.zip
12593045504/13510573713 [==========================>...] - ETA: 27s

# Get unique images
encode_train = sorted(set(img_name_vector))
# Feel free to change batch_size according to your system configuration
image_dataset = tf.data.Dataset.from_tensor_slices(encode_train)
image_dataset = image_dataset.map(
  load_image, num_parallel_calls=tf.data.experimental.AUTOTUNE).batch(16)
for img, path in image_dataset:
  batch_features = image_features_extract_model(img)
  batch_features = tf.reshape(batch_features,
                              (batch_features.shape[0], -1, batch_features.shape[3]))

  for bf, p in zip(batch_features, path):
    path_of_feature = p.numpy().decode("utf-8")
    np.save(path_of_feature, bf.numpy())

# Find the maximum length of any caption in our dataset
def calc_max_length(tensor):
    return max(len(t) for t in tensor)

# Choose the top 5000 words from the vocabulary
top_k = 5000
tokenizer = tf.keras.preprocessing.text.Tokenizer(num_words=top_k,
                                                  oov_token="<unk>",
                                                  filters='!"#$%&()*+.,-/:;=?@[\]^_`{|}~ ')
tokenizer.fit_on_texts(train_captions)
train_seqs = tokenizer.texts_to_sequences(train_captions)

tokenizer.word_index['<pad>'] = 0
tokenizer.index_word[0] = '<pad>'

# Create the tokenized vectors
train_seqs = tokenizer.texts_to_sequences(train_captions)

# Pad each vector to the max_length of the captions
# If you do not provide a max_length value, pad_sequences calculates it automatically
cap_vector = tf.keras.preprocessing.sequence.pad_sequences(train_seqs, padding='post')

# Calculates the max_length, which is used to store the attention weights
max_length = calc_max_length(train_seqs)

# Create training and validation sets using an 80-20 split
img_name_train, img_name_val, cap_train, cap_val = train_test_split(img_name_vector,
                                                                    cap_vector,
                                                                    test_size=0.2,
                                                                    random_state=0)

len(img_name_train), len(cap_train), len(img_name_val), len(cap_val)

(24000, 24000, 6000, 6000)

# Feel free to change these parameters according to your system's configuration

BATCH_SIZE = 64
BUFFER_SIZE = 1000
embedding_dim = 256
units = 512
vocab_size = len(tokenizer.word_index) + 1
num_steps = len(img_name_train) // BATCH_SIZE
# Shape of the vector extracted from InceptionV3 is (64, 2048)
# These two variables represent that vector shape
features_shape = 2048
attention_features_shape = 64

# Load the numpy files
def map_func(img_name, cap):
  img_tensor = np.load(img_name.decode('utf-8')+'.npy')
  return img_tensor, cap

dataset = tf.data.Dataset.from_tensor_slices((img_name_train, cap_train))
# Use map to load the numpy files in parallel
dataset = dataset.map(lambda item1, item2: tf.numpy_function(
          map_func, [item1, item2], [tf.float32, tf.int32]),
          num_parallel_calls=tf.data.experimental.AUTOTUNE)
# Shuffle and batch
dataset = dataset.shuffle(BUFFER_SIZE).batch(BATCH_SIZE)
dataset = dataset.prefetch(buffer_size=tf.data.experimental.AUTOTUNE)

class BahdanauAttention(tf.keras.Model):
  def __init__(self, units):
    super(BahdanauAttention, self).__init__()
    self.W1 = tf.keras.layers.Dense(units)
    self.W2 = tf.keras.layers.Dense(units)
    self.V = tf.keras.layers.Dense(1)

  def call(self, features, hidden):
    # features(CNN_encoder output) shape == (batch_size, 64, embedding_dim)

    # hidden shape == (batch_size, hidden_size)
    # hidden_with_time_axis shape == (batch_size, 1, hidden_size)
    hidden_with_time_axis = tf.expand_dims(hidden, 1)

    # score shape == (batch_size, 64, hidden_size)
    score = tf.nn.tanh(self.W1(features) + self.W2(hidden_with_time_axis))

    # attention_weights shape == (batch_size, 64, 1)
    # you get 1 at the last axis because you are applying score to self.V
    attention_weights = tf.nn.softmax(self.V(score), axis=1)

    # context_vector shape after sum == (batch_size, hidden_size)
    context_vector = attention_weights * features
    context_vector = tf.reduce_sum(context_vector, axis=1)

    return context_vector, attention_weights

class CNN_Encoder(tf.keras.Model):
    # Since you have already extracted the features and dumped it using pickle
    # This encoder passes those features through a Fully connected layer
    def __init__(self, embedding_dim):
        super(CNN_Encoder, self).__init__()
        # shape after fc == (batch_size, 64, embedding_dim)
        self.fc = tf.keras.layers.Dense(embedding_dim)

    def call(self, x):
        x = self.fc(x)
        x = tf.nn.relu(x)
        return x

class RNN_Decoder(tf.keras.Model):
  def __init__(self, embedding_dim, units, vocab_size):
    super(RNN_Decoder, self).__init__()
    self.units = units

    self.embedding = tf.keras.layers.Embedding(vocab_size, embedding_dim)
    self.gru = tf.keras.layers.GRU(self.units,
                                   return_sequences=True,
                                   return_state=True,
                                   recurrent_initializer='glorot_uniform')
    self.fc1 = tf.keras.layers.Dense(self.units)
    self.fc2 = tf.keras.layers.Dense(vocab_size)

    self.attention = BahdanauAttention(self.units)

  def call(self, x, features, hidden):
    # defining attention as a separate model
    context_vector, attention_weights = self.attention(features, hidden)

    # x shape after passing through embedding == (batch_size, 1, embedding_dim)
    x = self.embedding(x)

    # x shape after concatenation == (batch_size, 1, embedding_dim + hidden_size)
    x = tf.concat([tf.expand_dims(context_vector, 1), x], axis=-1)

    # passing the concatenated vector to the GRU
    output, state = self.gru(x)

    # shape == (batch_size, max_length, hidden_size)
    x = self.fc1(output)

    # x shape == (batch_size * max_length, hidden_size)
    x = tf.reshape(x, (-1, x.shape[2]))

    # output shape == (batch_size * max_length, vocab)
    x = self.fc2(x)

    return x, state, attention_weights

  def reset_state(self, batch_size):
    return tf.zeros((batch_size, self.units))

encoder = CNN_Encoder(embedding_dim)
decoder = RNN_Decoder(embedding_dim, units, vocab_size)

optimizer = tf.keras.optimizers.Adam()
loss_object = tf.keras.losses.SparseCategoricalCrossentropy(
    from_logits=True, reduction='none')
def loss_function(real, pred):
  mask = tf.math.logical_not(tf.math.equal(real, 0))
  loss_ = loss_object(real, pred)

  mask = tf.cast(mask, dtype=loss_.dtype)
  loss_ *= mask

  return tf.reduce_mean(loss_)

checkpoint_path = "./checkpoints/train"
ckpt = tf.train.Checkpoint(encoder=encoder,
                           decoder=decoder,
                           optimizer = optimizer)
ckpt_manager = tf.train.CheckpointManager(ckpt, checkpoint_path, max_to_keep=5)

start_epoch = 0
if ckpt_manager.latest_checkpoint:
  start_epoch = int(ckpt_manager.latest_checkpoint.split('-')[-1])

# adding this in a separate cell because if you run the training cell
# many times, the loss_plot array will be reset
loss_plot = []

@tf.function
def train_step(img_tensor, target):
  loss = 0

  # initializing the hidden state for each batch
  # because the captions are not related from image to image
  hidden = decoder.reset_state(batch_size=target.shape[0])

  dec_input = tf.expand_dims([tokenizer.word_index['<start>']] * BATCH_SIZE, 1)

  with tf.GradientTape() as tape:
      features = encoder(img_tensor)

      for i in range(1, target.shape[1]):
          # passing the features through the decoder
          predictions, hidden, _ = decoder(dec_input, features, hidden)

          loss += loss_function(target[:, i], predictions)

          # using teacher forcing
          dec_input = tf.expand_dims(target[:, i], 1)

  total_loss = (loss / int(target.shape[1]))

  trainable_variables = encoder.trainable_variables + decoder.trainable_variables

  gradients = tape.gradient(loss, trainable_variables)

  optimizer.apply_gradients(zip(gradients, trainable_variables))

  return loss, total_loss

EPOCHS = 20
for epoch in range(start_epoch, EPOCHS):
    start = time.time()
    total_loss = 0

    for (batch, (img_tensor, target)) in enumerate(dataset):
        batch_loss, t_loss = train_step(img_tensor, target)
        total_loss += t_loss

        if batch % 100 == 0:
            print ('Epoch {} Batch {} Loss {:.4f}'.format(
              epoch + 1, batch, batch_loss.numpy() / int(target.shape[1])))
    # storing the epoch end loss value to plot later
    loss_plot.append(total_loss / num_steps)

    if epoch % 5 == 0:
      ckpt_manager.save()

    print ('Epoch {} Loss {:.6f}'.format(epoch + 1,
                                         total_loss/num_steps))
    print ('Time taken for 1 epoch {} sec\n'.format(time.time() - start))

Epoch 1 Batch 0 Loss 2.0988
Epoch 1 Batch 100 Loss 1.1463
Epoch 1 Batch 200 Loss 1.0366
Epoch 1 Batch 300 Loss 0.9083
Epoch 1 Loss 1.085102
Time taken for 1 epoch 131.9147825241089 sec

Epoch 2 Batch 0 Loss 0.8748
Epoch 2 Batch 100 Loss 0.7652
Epoch 2 Batch 200 Loss 0.7708
Epoch 2 Batch 300 Loss 0.7578
Epoch 2 Loss 0.812931
Time taken for 1 epoch 49.82423710823059 sec

Epoch 3 Batch 0 Loss 0.8118
Epoch 3 Batch 100 Loss 0.7946
Epoch 3 Batch 200 Loss 0.7396
Epoch 3 Batch 300 Loss 0.6746
Epoch 3 Loss 0.735972
Time taken for 1 epoch 49.87708878517151 sec

Epoch 4 Batch 0 Loss 0.6586
Epoch 4 Batch 100 Loss 0.7100
Epoch 4 Batch 200 Loss 0.6617
Epoch 4 Batch 300 Loss 0.7083
Epoch 4 Loss 0.688551
Time taken for 1 epoch 49.925899028778076 sec

Epoch 5 Batch 0 Loss 0.6543
Epoch 5 Batch 100 Loss 0.6995
Epoch 5 Batch 200 Loss 0.6268
Epoch 5 Batch 300 Loss 0.6577
Epoch 5 Loss 0.650644
Time taken for 1 epoch 49.861159801483154 sec

Epoch 6 Batch 0 Loss 0.6031
Epoch 6 Batch 100 Loss 0.5955
Epoch 6 Batch 200 Loss 0.6627
Epoch 6 Batch 300 Loss 0.5704
Epoch 6 Loss 0.617129
Time taken for 1 epoch 50.251293659210205 sec

Epoch 7 Batch 0 Loss 0.5712
Epoch 7 Batch 100 Loss 0.5685
Epoch 7 Batch 200 Loss 0.5779
Epoch 7 Batch 300 Loss 0.5544
Epoch 7 Loss 0.586807
Time taken for 1 epoch 50.4225971698761 sec

Epoch 8 Batch 0 Loss 0.5429
Epoch 8 Batch 100 Loss 0.5585
Epoch 8 Batch 200 Loss 0.5514
Epoch 8 Batch 300 Loss 0.5229
Epoch 8 Loss 0.555478
Time taken for 1 epoch 50.04306435585022 sec

Epoch 9 Batch 0 Loss 0.5150
Epoch 9 Batch 100 Loss 0.5001
Epoch 9 Batch 200 Loss 0.5294
Epoch 9 Batch 300 Loss 0.5434
Epoch 9 Loss 0.526150
Time taken for 1 epoch 49.535995960235596 sec

Epoch 10 Batch 0 Loss 0.4677
Epoch 10 Batch 100 Loss 0.5044
Epoch 10 Batch 200 Loss 0.4583
Epoch 10 Batch 300 Loss 0.4794
Epoch 10 Loss 0.496457
Time taken for 1 epoch 50.2047655582428 sec

Epoch 11 Batch 0 Loss 0.4150
Epoch 11 Batch 100 Loss 0.4491
Epoch 11 Batch 200 Loss 0.4283
Epoch 11 Batch 300 Loss 0.4874
Epoch 11 Loss 0.465688
Time taken for 1 epoch 50.450185775756836 sec

Epoch 12 Batch 0 Loss 0.4305
Epoch 12 Batch 100 Loss 0.4535
Epoch 12 Batch 200 Loss 0.4198
Epoch 12 Batch 300 Loss 0.4154
Epoch 12 Loss 0.437214
Time taken for 1 epoch 49.61044931411743 sec

Epoch 13 Batch 0 Loss 0.4156
Epoch 13 Batch 100 Loss 0.4067
Epoch 13 Batch 200 Loss 0.4412
Epoch 13 Batch 300 Loss 0.4066
Epoch 13 Loss 0.429518
Time taken for 1 epoch 50.13954949378967 sec

Epoch 14 Batch 0 Loss 0.3823
Epoch 14 Batch 100 Loss 0.4156
Epoch 14 Batch 200 Loss 0.3560
Epoch 14 Batch 300 Loss 0.4084
Epoch 14 Loss 0.387618
Time taken for 1 epoch 49.05424618721008 sec

Epoch 15 Batch 0 Loss 0.3724
Epoch 15 Batch 100 Loss 0.3452
Epoch 15 Batch 200 Loss 0.3371
Epoch 15 Batch 300 Loss 0.3183
Epoch 15 Loss 0.358968
Time taken for 1 epoch 49.87037777900696 sec

Epoch 16 Batch 0 Loss 0.3415
Epoch 16 Batch 100 Loss 0.3094
Epoch 16 Batch 200 Loss 0.3534
Epoch 16 Batch 300 Loss 0.3220
Epoch 16 Loss 0.340680
Time taken for 1 epoch 50.09799098968506 sec

Epoch 17 Batch 0 Loss 0.3501
Epoch 17 Batch 100 Loss 0.3355
Epoch 17 Batch 200 Loss 0.3027
Epoch 17 Batch 300 Loss 0.3440
Epoch 17 Loss 0.318385
Time taken for 1 epoch 49.605764865875244 sec

Epoch 18 Batch 0 Loss 0.3254
Epoch 18 Batch 100 Loss 0.3095
Epoch 18 Batch 200 Loss 0.2968
Epoch 18 Batch 300 Loss 0.2670
Epoch 18 Loss 0.295994
Time taken for 1 epoch 49.70194149017334 sec

Epoch 19 Batch 0 Loss 0.3094
Epoch 19 Batch 100 Loss 0.3093
Epoch 19 Batch 200 Loss 0.2804
Epoch 19 Batch 300 Loss 0.2976
Epoch 19 Loss 0.278130
Time taken for 1 epoch 49.86575794219971 sec

Epoch 20 Batch 0 Loss 0.2911
Epoch 20 Batch 100 Loss 0.2470
Epoch 20 Batch 200 Loss 0.2651
Epoch 20 Batch 300 Loss 0.2656
Epoch 20 Loss 0.258760
Time taken for 1 epoch 50.28017234802246 sec

plt.plot(loss_plot)
plt.xlabel('Epochs')
plt.ylabel('Loss')
plt.title('Loss Plot')
plt.show()

def evaluate(image):
    attention_plot = np.zeros((max_length, attention_features_shape))

    hidden = decoder.reset_state(batch_size=1)

    temp_input = tf.expand_dims(load_image(image)[0], 0)
    img_tensor_val = image_features_extract_model(temp_input)
    img_tensor_val = tf.reshape(img_tensor_val, (img_tensor_val.shape[0], -1, img_tensor_val.shape[3]))

    features = encoder(img_tensor_val)

    dec_input = tf.expand_dims([tokenizer.word_index['<start>']], 0)
    result = []

    for i in range(max_length):
        predictions, hidden, attention_weights = decoder(dec_input, features, hidden)

        attention_plot[i] = tf.reshape(attention_weights, (-1, )).numpy()

        predicted_id = tf.argmax(predictions[0]).numpy()
        result.append(tokenizer.index_word[predicted_id])

        if tokenizer.index_word[predicted_id] == '<end>':
            return result, attention_plot

        dec_input = tf.expand_dims([predicted_id], 0)

    attention_plot = attention_plot[:len(result), :]
    return result, attention_plot

def plot_attention(image, result, attention_plot):
    temp_image = np.array(Image.open(image))

    fig = plt.figure(figsize=(10, 10))

    len_result = len(result)
    for l in range(len_result):
        temp_att = np.resize(attention_plot[l], (8, 8))
        ax = fig.add_subplot(len_result//2, len_result//2, l+1)
        ax.set_title(result[l])
        img = ax.imshow(temp_image)
        ax.imshow(temp_att, cmap='gray', alpha=0.6, extent=img.get_extent())

    plt.tight_layout()
    plt.show()

# captions on the validation set
rid = np.random.randint(0, len(img_name_val))
image = img_name_val[rid]
real_caption = ' '.join([tokenizer.index_word[i] for i in cap_val[rid] if i not in [0]])
result, attention_plot = evaluate(image)
print ('Real Caption:', real_caption)
print ('Prediction Caption:', ' '.join(result))
plot_attention(image, result, attention_plot)
# opening the image
Image.open(img_name_val[rid])

Real Caption: <start> a bike parked next to a bucket filled with lots of oranges <end>
Prediction Caption: a man sitting in the banana <end>

image_url = 'https://tensorflow.org/images/surf.jpg'
image_extension = image_url[-4:]
image_path = tf.keras.utils.get_file('image'+image_extension,
                                     origin=image_url)

result, attention_plot = evaluate(image_path)
print ('Prediction Caption:', ' '.join(result))
plot_attention(image_path, result, attention_plot)
# opening the image
Image.open(image_path)

Downloading data from https://tensorflow.org/images/surf.jpg
65536/64400 [==============================] - 0s 2us/step
Prediction Caption: a man on a surfboard riding a surfboard <end>