ML Community Day is November 9! Join us for updates from TensorFlow, JAX, and more Learn more

Image captioning with visual attention

View on TensorFlow.org View source on GitHub Download notebook

Given an image like the example below, your goal is to generate a caption such as "a surfer riding on a wave".

Man Surfing

Image Source; License: Public Domain

To accomplish this, you'll use an attention-based model, which enables us to see what parts of the image the model focuses on as it generates a caption.

Prediction

The model architecture is similar to Show, Attend and Tell: Neural Image Caption Generation with Visual Attention.

This notebook is an end-to-end example. When you run the notebook, it downloads the MS-COCO dataset, preprocesses and caches a subset of images using Inception V3, trains an encoder-decoder model, and generates captions on new images using the trained model.

In this example, you will train a model on a relatively small amount of data—the first 30,000 captions for about 20,000 images (because there are multiple captions per image in the dataset).

import tensorflow as tf

# You'll generate plots of attention in order to see which parts of an image
# your model focuses on during captioning
import matplotlib.pyplot as plt

import collections
import random
import numpy as np
import os
import time
import json
from PIL import Image

2021-07-28 01:23:48.839502: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcudart.so.11.0

Download and prepare the MS-COCO dataset

You will use the MS-COCO dataset to train your model. The dataset contains over 82,000 images, each of which has at least 5 different caption annotations. The code below downloads and extracts the dataset automatically.

# Download caption annotation files
annotation_folder = '/annotations/'
if not os.path.exists(os.path.abspath('.') + annotation_folder):
  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'
  os.remove(annotation_zip)

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

Downloading data from http://images.cocodataset.org/annotations/annotations_trainval2014.zip
252878848/252872794 [==============================] - 17s 0us/step
Downloading data from http://images.cocodataset.org/zips/train2014.zip
13510574080/13510573713 [==============================] - 816s 0us/step

Optional: limit the size of the training set

To speed up training for this tutorial, you'll use a subset of 30,000 captions and their corresponding images to train your model. Choosing to use more data would result in improved captioning quality.

with open(annotation_file, 'r') as f:
    annotations = json.load(f)

# Group all captions together having the same image ID.
image_path_to_caption = collections.defaultdict(list)
for val in annotations['annotations']:
  caption = f"<start> {val['caption']} <end>"
  image_path = PATH + 'COCO_train2014_' + '%012d.jpg' % (val['image_id'])
  image_path_to_caption[image_path].append(caption)

image_paths = list(image_path_to_caption.keys())
random.shuffle(image_paths)

# Select the first 6000 image_paths from the shuffled set.
# Approximately each image id has 5 captions associated with it, so that will
# lead to 30,000 examples.
train_image_paths = image_paths[:6000]
print(len(train_image_paths))

6000

train_captions = []
img_name_vector = []

for image_path in train_image_paths:
  caption_list = image_path_to_caption[image_path]
  train_captions.extend(caption_list)
  img_name_vector.extend([image_path] * len(caption_list))

print(train_captions[0])
Image.open(img_name_vector[0])

<start> a close up of a person wearing a bow tie  <end>

png

Preprocess the images using InceptionV3

Next, you will use InceptionV3 (which is pretrained on Imagenet) to classify each image. You will extract features from the last convolutional layer.

First, you will convert the images into InceptionV3's expected format by:

  • Resizing the image to 299px by 299px
  • Preprocess the images using the preprocess_input method to normalize the image so that it contains pixels in the range of -1 to 1, which matches the format of the images used to train InceptionV3.
def load_image(image_path):
    img = tf.io.read_file(image_path)
    img = tf.image.decode_jpeg(img, channels=3)
    img = tf.image.resize(img, (299, 299))
    img = tf.keras.applications.inception_v3.preprocess_input(img)
    return img, image_path

Initialize InceptionV3 and load the pretrained Imagenet weights

Now you'll create a tf.keras model where the output layer is the last convolutional layer in the InceptionV3 architecture. The shape of the output of this layer is 8x8x2048. You use the last convolutional layer because you are using attention in this example. You don't perform this initialization during training because it could become a bottleneck.

  • You forward each image through the network and store the resulting vector in a dictionary (image_name --> feature_vector).
  • After all the images are passed through the network, you save the dictionary to disk.
image_model = tf.keras.applications.InceptionV3(include_top=False,
                                                weights='imagenet')
new_input = image_model.input
hidden_layer = image_model.layers[-1].output

image_features_extract_model = tf.keras.Model(new_input, hidden_layer)

2021-07-28 01:38:51.724662: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcuda.so.1
2021-07-28 01:38:52.396689: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2021-07-28 01:38:52.397733: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1733] Found device 0 with properties: 
pciBusID: 0000:00:05.0 name: Tesla V100-SXM2-16GB computeCapability: 7.0
coreClock: 1.53GHz coreCount: 80 deviceMemorySize: 15.78GiB deviceMemoryBandwidth: 836.37GiB/s
2021-07-28 01:38:52.397773: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcudart.so.11.0
2021-07-28 01:38:52.410682: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcublas.so.11
2021-07-28 01:38:52.410775: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcublasLt.so.11
2021-07-28 01:38:52.412998: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcufft.so.10
2021-07-28 01:38:52.416058: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcurand.so.10
2021-07-28 01:38:52.418038: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcusolver.so.11
2021-07-28 01:38:52.420886: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcusparse.so.11
2021-07-28 01:38:52.421795: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcudnn.so.8
2021-07-28 01:38:52.421939: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2021-07-28 01:38:52.422953: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2021-07-28 01:38:52.423807: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1871] Adding visible gpu devices: 0
2021-07-28 01:38:52.424745: I tensorflow/core/platform/cpu_feature_guard.cc:142] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  AVX2 AVX512F FMA
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.
2021-07-28 01:38:52.425332: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2021-07-28 01:38:52.426304: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1733] Found device 0 with properties: 
pciBusID: 0000:00:05.0 name: Tesla V100-SXM2-16GB computeCapability: 7.0
coreClock: 1.53GHz coreCount: 80 deviceMemorySize: 15.78GiB deviceMemoryBandwidth: 836.37GiB/s
2021-07-28 01:38:52.426381: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2021-07-28 01:38:52.427321: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2021-07-28 01:38:52.428215: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1871] Adding visible gpu devices: 0
2021-07-28 01:38:52.429333: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcudart.so.11.0
2021-07-28 01:38:53.947925: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1258] Device interconnect StreamExecutor with strength 1 edge matrix:
2021-07-28 01:38:53.947964: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1264]      0 
2021-07-28 01:38:53.947973: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1277] 0:   N 
2021-07-28 01:38:53.949235: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2021-07-28 01:38:53.950377: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2021-07-28 01:38:53.951333: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:937] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2021-07-28 01:38:53.952280: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1418] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 14646 MB memory) -> physical GPU (device: 0, name: Tesla V100-SXM2-16GB, pci bus id: 0000:00:05.0, compute capability: 7.0)
Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/inception_v3/inception_v3_weights_tf_dim_ordering_tf_kernels_notop.h5
87916544/87910968 [==============================] - 4s 0us/step

Caching the features extracted from InceptionV3

You will pre-process each image with InceptionV3 and cache the output to disk. Caching the output in RAM would be faster but also memory intensive, requiring 8 * 8 * 2048 floats per image. At the time of writing, this exceeds the memory limitations of Colab (currently 12GB of memory).

Performance could be improved with a more sophisticated caching strategy (for example, by sharding the images to reduce random access disk I/O), but that would require more code.

The caching will take about 10 minutes to run in Colab with a GPU. If you'd like to see a progress bar, you can:

  1. Install tqdm:

    !pip install tqdm

  2. Import tqdm:

    from tqdm import tqdm

  3. Change the following line:

    for img, path in image_dataset:

    to:

    for img, path in tqdm(image_dataset):

# 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.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())

2021-07-28 01:39:00.906130: I tensorflow/compiler/mlir/mlir_graph_optimization_pass.cc:176] None of the MLIR Optimization Passes are enabled (registered 2)
2021-07-28 01:39:00.907617: I tensorflow/core/platform/profile_utils/cpu_utils.cc:114] CPU Frequency: 2000179999 Hz
2021-07-28 01:39:01.056105: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcudnn.so.8
2021-07-28 01:39:03.207340: I tensorflow/stream_executor/cuda/cuda_dnn.cc:359] Loaded cuDNN version 8100
2021-07-28 01:39:08.876833: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcublas.so.11
2021-07-28 01:39:10.057729: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcublasLt.so.11

Preprocess and tokenize the captions

  • First, you'll tokenize the captions (for example, by splitting on spaces). This gives us a vocabulary of all of the unique words in the data (for example, "surfing", "football", and so on).
  • Next, you'll limit the vocabulary size to the top 5,000 words (to save memory). You'll replace all other words with the token "UNK" (unknown).
  • You then create word-to-index and index-to-word mappings.
  • Finally, you pad all sequences to be the same length as the longest one.
# Find the maximum length of any caption in the 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)

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)

Split the data into training and testing

img_to_cap_vector = collections.defaultdict(list)
for img, cap in zip(img_name_vector, cap_vector):
  img_to_cap_vector[img].append(cap)

# Create training and validation sets using an 80-20 split randomly.
img_keys = list(img_to_cap_vector.keys())
random.shuffle(img_keys)

slice_index = int(len(img_keys)*0.8)
img_name_train_keys, img_name_val_keys = img_keys[:slice_index], img_keys[slice_index:]

img_name_train = []
cap_train = []
for imgt in img_name_train_keys:
  capt_len = len(img_to_cap_vector[imgt])
  img_name_train.extend([imgt] * capt_len)
  cap_train.extend(img_to_cap_vector[imgt])

img_name_val = []
cap_val = []
for imgv in img_name_val_keys:
  capv_len = len(img_to_cap_vector[imgv])
  img_name_val.extend([imgv] * capv_len)
  cap_val.extend(img_to_cap_vector[imgv])

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

(24012, 24012, 6007, 6007)

Create a tf.data dataset for training

Your images and captions are ready! Next, let's create a tf.data dataset to use for training your model.

# 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 = top_k + 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.AUTOTUNE)

# Shuffle and batch
dataset = dataset.shuffle(BUFFER_SIZE).batch(BATCH_SIZE)
dataset = dataset.prefetch(buffer_size=tf.data.AUTOTUNE)

Model

Fun fact: the decoder below is identical to the one in the example for Neural Machine Translation with Attention.

The model architecture is inspired by the Show, Attend and Tell paper.

  • In this example, you extract the features from the lower convolutional layer of InceptionV3 giving us a vector of shape (8, 8, 2048).
  • You squash that to a shape of (64, 2048).
  • This vector is then passed through the CNN Encoder (which consists of a single Fully connected layer).
  • The RNN (here GRU) attends over the image to predict the next word.
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)

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

    # score shape == (batch_size, 64, 1)
    # This gives you an unnormalized score for each image feature.
    score = self.V(attention_hidden_layer)

    # attention_weights shape == (batch_size, 64, 1)
    attention_weights = tf.nn.softmax(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
    # 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

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])
  # restoring the latest checkpoint in checkpoint_path
  ckpt.restore(ckpt_manager.latest_checkpoint)

Training

  • You extract the features stored in the respective .npy files and then pass those features through the encoder.
  • The encoder output, hidden state(initialized to 0) and the decoder input (which is the start token) is passed to the decoder.
  • The decoder returns the predictions and the decoder hidden state.
  • The decoder hidden state is then passed back into the model and the predictions are used to calculate the loss.
  • Use teacher forcing to decide the next input to the decoder.
  • Teacher forcing is the technique where the target word is passed as the next input to the decoder.
  • The final step is to calculate the gradients and apply it to the optimizer and backpropagate.
# 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>']] * target.shape[0], 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:
            average_batch_loss = batch_loss.numpy()/int(target.shape[1])
            print(f'Epoch {epoch+1} Batch {batch} Loss {average_batch_loss:.4f}')
    # storing the epoch end loss value to plot later
    loss_plot.append(total_loss / num_steps)

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

    print(f'Epoch {epoch+1} Loss {total_loss/num_steps:.6f}')
    print(f'Time taken for 1 epoch {time.time()-start:.2f} sec\n')

Epoch 1 Batch 0 Loss 1.9004
Epoch 1 Batch 100 Loss 1.0669
Epoch 1 Batch 200 Loss 0.8644
Epoch 1 Batch 300 Loss 0.7575
Epoch 1 Loss 0.971214
Time taken for 1 epoch 150.00 sec

Epoch 2 Batch 0 Loss 0.8469
Epoch 2 Batch 100 Loss 0.7256
Epoch 2 Batch 200 Loss 0.7352
Epoch 2 Batch 300 Loss 0.6788
Epoch 2 Loss 0.740390
Time taken for 1 epoch 52.59 sec

Epoch 3 Batch 0 Loss 0.7644
Epoch 3 Batch 100 Loss 0.6992
Epoch 3 Batch 200 Loss 0.6509
Epoch 3 Batch 300 Loss 0.5881
Epoch 3 Loss 0.665821
Time taken for 1 epoch 52.17 sec

Epoch 4 Batch 0 Loss 0.6438
Epoch 4 Batch 100 Loss 0.5957
Epoch 4 Batch 200 Loss 0.6577
Epoch 4 Batch 300 Loss 0.6111
Epoch 4 Loss 0.617179
Time taken for 1 epoch 51.69 sec

Epoch 5 Batch 0 Loss 0.5838
Epoch 5 Batch 100 Loss 0.6093
Epoch 5 Batch 200 Loss 0.6297
Epoch 5 Batch 300 Loss 0.5459
Epoch 5 Loss 0.579694
Time taken for 1 epoch 51.83 sec

Epoch 6 Batch 0 Loss 0.5737
Epoch 6 Batch 100 Loss 0.5443
Epoch 6 Batch 200 Loss 0.5537
Epoch 6 Batch 300 Loss 0.5825
Epoch 6 Loss 0.546888
Time taken for 1 epoch 51.11 sec

Epoch 7 Batch 0 Loss 0.5380
Epoch 7 Batch 100 Loss 0.5383
Epoch 7 Batch 200 Loss 0.4824
Epoch 7 Batch 300 Loss 0.4913
Epoch 7 Loss 0.517661
Time taken for 1 epoch 50.39 sec

Epoch 8 Batch 0 Loss 0.4940
Epoch 8 Batch 100 Loss 0.5097
Epoch 8 Batch 200 Loss 0.5023
Epoch 8 Batch 300 Loss 0.4532
Epoch 8 Loss 0.490831
Time taken for 1 epoch 50.93 sec

Epoch 9 Batch 0 Loss 0.4901
Epoch 9 Batch 100 Loss 0.4050
Epoch 9 Batch 200 Loss 0.4870
Epoch 9 Batch 300 Loss 0.4596
Epoch 9 Loss 0.465417
Time taken for 1 epoch 51.03 sec

Epoch 10 Batch 0 Loss 0.4536
Epoch 10 Batch 100 Loss 0.4588
Epoch 10 Batch 200 Loss 0.4160
Epoch 10 Batch 300 Loss 0.4242
Epoch 10 Loss 0.440437
Time taken for 1 epoch 51.10 sec

Epoch 11 Batch 0 Loss 0.4445
Epoch 11 Batch 100 Loss 0.4225
Epoch 11 Batch 200 Loss 0.4142
Epoch 11 Batch 300 Loss 0.4043
Epoch 11 Loss 0.418332
Time taken for 1 epoch 51.65 sec

Epoch 12 Batch 0 Loss 0.4569
Epoch 12 Batch 100 Loss 0.3960
Epoch 12 Batch 200 Loss 0.3994
Epoch 12 Batch 300 Loss 0.3606
Epoch 12 Loss 0.396864
Time taken for 1 epoch 50.60 sec

Epoch 13 Batch 0 Loss 0.3756
Epoch 13 Batch 100 Loss 0.3739
Epoch 13 Batch 200 Loss 0.3485
Epoch 13 Batch 300 Loss 0.3188
Epoch 13 Loss 0.377340
Time taken for 1 epoch 50.36 sec

Epoch 14 Batch 0 Loss 0.3615
Epoch 14 Batch 100 Loss 0.3441
Epoch 14 Batch 200 Loss 0.3526
Epoch 14 Batch 300 Loss 0.3481
Epoch 14 Loss 0.357066
Time taken for 1 epoch 50.84 sec

Epoch 15 Batch 0 Loss 0.3696
Epoch 15 Batch 100 Loss 0.3506
Epoch 15 Batch 200 Loss 0.3470
Epoch 15 Batch 300 Loss 0.3242
Epoch 15 Loss 0.339348
Time taken for 1 epoch 50.10 sec

Epoch 16 Batch 0 Loss 0.3250
Epoch 16 Batch 100 Loss 0.3281
Epoch 16 Batch 200 Loss 0.3296
Epoch 16 Batch 300 Loss 0.3140
Epoch 16 Loss 0.321988
Time taken for 1 epoch 50.75 sec

Epoch 17 Batch 0 Loss 0.2916
Epoch 17 Batch 100 Loss 0.2957
Epoch 17 Batch 200 Loss 0.3014
Epoch 17 Batch 300 Loss 0.2942
Epoch 17 Loss 0.306097
Time taken for 1 epoch 51.17 sec

Epoch 18 Batch 0 Loss 0.2839
Epoch 18 Batch 100 Loss 0.2937
Epoch 18 Batch 200 Loss 0.2837
Epoch 18 Batch 300 Loss 0.2717
Epoch 18 Loss 0.291026
Time taken for 1 epoch 49.86 sec

Epoch 19 Batch 0 Loss 0.3187
Epoch 19 Batch 100 Loss 0.3106
Epoch 19 Batch 200 Loss 0.2696
Epoch 19 Batch 300 Loss 0.2765
Epoch 19 Loss 0.278087
Time taken for 1 epoch 51.39 sec

Epoch 20 Batch 0 Loss 0.2823
Epoch 20 Batch 100 Loss 0.2791
Epoch 20 Batch 200 Loss 0.2556
Epoch 20 Batch 300 Loss 0.2598
Epoch 20 Loss 0.264479
Time taken for 1 epoch 51.29 sec

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

png

Caption!

  • The evaluate function is similar to the training loop, except you don't use teacher forcing here. The input to the decoder at each time step is its previous predictions along with the hidden state and the encoder output.
  • Stop predicting when the model predicts the end token.
  • And store the attention weights for every time step.
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.random.categorical(predictions, 1)[0][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 i in range(len_result):
        temp_att = np.resize(attention_plot[i], (8, 8))
        grid_size = max(np.ceil(len_result/2), 2)
        ax = fig.add_subplot(grid_size, grid_size, i+1)
        ax.set_title(result[i])
        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)

Real Caption: <start> some people are playing a game in a field <end>
Prediction Caption: a man and little boy that is playing frisbee in a <unk> <end>
/home/kbuilder/.local/lib/python3.7/site-packages/ipykernel_launcher.py:10: MatplotlibDeprecationWarning: Passing non-integers as three-element position specification is deprecated since 3.3 and will be removed two minor releases later.
  # Remove the CWD from sys.path while we load stuff.

png

Try it on your own images

For fun, below you're provided a method you can use to caption your own images with the model you've just trained. Keep in mind, it was trained on a relatively small amount of data, and your images may be different from the training data (so be prepared for weird results!)

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 5us/step
Prediction Caption: a man in <unk> as he rides a surf board <end>
/home/kbuilder/.local/lib/python3.7/site-packages/ipykernel_launcher.py:10: MatplotlibDeprecationWarning: Passing non-integers as three-element position specification is deprecated since 3.3 and will be removed two minor releases later.
  # Remove the CWD from sys.path while we load stuff.

png

png

Next steps

Congrats! You've just trained an image captioning model with attention. Next, take a look at this example Neural Machine Translation with Attention. It uses a similar architecture to translate between Spanish and English sentences. You can also experiment with training the code in this notebook on a different dataset.