Image Caption Generation
Image from paper Show, Attend and TellProject 71: Automatic Image Caption Generation
Authors: Oihane Cantero and Julen Etxaniz
Supervisors: Oier Lopez de Lacalle and Eneko Agirre
Subject: Machine Learning and Neural Networks
Date: 20-12-2020
Objective: Implement a caption generation model that uses a CNN to condition a LSTM based language model.
Contents:
1.Import Libraries
2.Get Dataset
3.Prepare Photo Data
4.Prepare Text Data
5.Load Data
6.Encode Text Data
7.Define Model
8.Fit Model
9.Evaluate Model
10.Generate Captions
References
[1] https://arxiv.org/pdf/1411.4555.pdf
[2] https://machinelearningmastery.com/develop-a-deep-learning-caption-generation-model-in-python/
[3] https://github.com/dabasajay/Image-Caption-Generator
[4] https://github.com/yashk2810/Image-Captioning
1. Import Libraries
# Prepare Photo Data
from os import listdir
from os.path import isfile
from pickle import dump
from tqdm import tqdm
from keras.models import Model
from keras.applications.vgg16 import VGG16
from keras.applications.inception_v3 import InceptionV3
from keras.preprocessing.image import load_img
from keras.preprocessing.image import img_to_array
import matplotlib.pyplot as plt
# Prepare Text Data
import string
# Load Data
from pickle import load
# Encode Text Data
from keras.preprocessing.text import Tokenizer
# Define Model
from keras.utils import plot_model
from keras.models import Model
from keras.models import load_model
from keras.layers import Input
from keras.layers import Dense
from keras.layers import LSTM
from keras.layers import Embedding
from keras.layers import Dropout
from keras.layers.merge import add
from keras.layers import RepeatVector, TimeDistributed, concatenate, Bidirectional
# Fit Model
import numpy as np
from keras.preprocessing.sequence import pad_sequences
from keras.utils import to_categorical
# Evaluate Model
from numpy import argmax, argsort
from nltk.translate.bleu_score import corpus_bleu, sentence_bleu, SmoothingFunction
# Generate Captions
from IPython.display import Image, display
2. Get Dataset
We decided to use Flickr8k Dataset. It can be downloaded from: https://github.com/jbrownlee/Datasets/releases/tag/Flickr8k
It has 8092 images and 5 captions for each image. Each image has 5 captions because obviously, there are different ways to caption an image. This dataset has predefined training, testing and evaluation subsets of 6000, 1000 and 1000 images respectively.
3. Prepare Photo Data
Two different models to extract image features: VGG16 and InceptionV3.
# preprocess the image for the model
def preprocess_image(filename, image_size):
image = load_img(filename, target_size=(image_size, image_size))
# convert the image pixels to a numpy array
image = img_to_array(image)
# reshape data for the model
image = image.reshape((1, image.shape[0], image.shape[1], image.shape[2]))
# prepare the image for the model
image = preprocess_input(image)
return image
# extract features from each photo in the directory
def extract_features(directory, model, image_size):
# extract features from each photo
features = dict()
for name in tqdm(listdir(directory), position=0, leave=True):
# load an image from file
filename = directory + '/' + name
# preprocess the image for the model
image = preprocess_image(filename, image_size)
# get features
feature = model.predict(image, verbose=0)
# get image id
image_id = name.split('.')[0]
# store feature
features[image_id] = feature
return features
def plot(model, filename):
plot_model(model, to_file=filename, show_shapes=True, show_layer_names=False)
display(Image(filename))
3.1. VGG16 Model
from keras.applications.vgg16 import preprocess_input
# cnn VGG16 model
def cnn_vgg16():
# load the model
model = VGG16()
# re-structure the model
model = Model(inputs=model.inputs, outputs=model.layers[-2].output)
# summarize
print(model.summary())
return model
# get cnn vgg16 model
cnn_vgg16 = cnn_vgg16()
Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/vgg16/vgg16_weights_tf_dim_ordering_tf_kernels.h5
553467904/553467096 [==============================] - 4s 0us/step
Model: "model"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
input_1 (InputLayer) [(None, 224, 224, 3)] 0
_________________________________________________________________
block1_conv1 (Conv2D) (None, 224, 224, 64) 1792
_________________________________________________________________
block1_conv2 (Conv2D) (None, 224, 224, 64) 36928
_________________________________________________________________
block1_pool (MaxPooling2D) (None, 112, 112, 64) 0
_________________________________________________________________
block2_conv1 (Conv2D) (None, 112, 112, 128) 73856
_________________________________________________________________
block2_conv2 (Conv2D) (None, 112, 112, 128) 147584
_________________________________________________________________
block2_pool (MaxPooling2D) (None, 56, 56, 128) 0
_________________________________________________________________
block3_conv1 (Conv2D) (None, 56, 56, 256) 295168
_________________________________________________________________
block3_conv2 (Conv2D) (None, 56, 56, 256) 590080
_________________________________________________________________
block3_conv3 (Conv2D) (None, 56, 56, 256) 590080
_________________________________________________________________
block3_pool (MaxPooling2D) (None, 28, 28, 256) 0
_________________________________________________________________
block4_conv1 (Conv2D) (None, 28, 28, 512) 1180160
_________________________________________________________________
block4_conv2 (Conv2D) (None, 28, 28, 512) 2359808
_________________________________________________________________
block4_conv3 (Conv2D) (None, 28, 28, 512) 2359808
_________________________________________________________________
block4_pool (MaxPooling2D) (None, 14, 14, 512) 0
_________________________________________________________________
block5_conv1 (Conv2D) (None, 14, 14, 512) 2359808
_________________________________________________________________
block5_conv2 (Conv2D) (None, 14, 14, 512) 2359808
_________________________________________________________________
block5_conv3 (Conv2D) (None, 14, 14, 512) 2359808
_________________________________________________________________
block5_pool (MaxPooling2D) (None, 7, 7, 512) 0
_________________________________________________________________
flatten (Flatten) (None, 25088) 0
_________________________________________________________________
fc1 (Dense) (None, 4096) 102764544
_________________________________________________________________
fc2 (Dense) (None, 4096) 16781312
=================================================================
Total params: 134,260,544
Trainable params: 134,260,544
Non-trainable params: 0
_________________________________________________________________
None
# plot model
filename = 'models/cnn_vgg16.png'
plot(cnn_vgg16, filename)
filename = 'files/features_vgg16.pkl'
# only extract if file does not exist
if not isfile(filename):
# extract features from all images
directory = 'Flickr8k_Dataset'
features = extract_features(directory, cnn_vgg16, 224)
# save to file
dump(features, open(filename, 'wb'))
# display original and preprocessed image
example_image = "Flickr8k_Dataset/667626_18933d713e.jpg"
display(Image(example_image))
image = preprocess_image(example_image, 224)
plt.imshow(np.squeeze(image))
3.2. InceptionV3 Model
from keras.applications.inception_v3 import preprocess_input
# cnn InceptionV3 model
def cnn_inceptionv3():
# load the model
model = InceptionV3(weights='imagenet')
# re-structure the model
model = Model(inputs=model.inputs, outputs=model.layers[-2].output)
# summarize
print(model.summary())
return model
# get cnn inceptionv3 model
cnn_inceptionv3 = cnn_inceptionv3()
Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/inception_v3/inception_v3_weights_tf_dim_ordering_tf_kernels.h5
96116736/96112376 [==============================] - 0s 0us/step
Model: "model"
__________________________________________________________________________________________________
Layer (type) Output Shape Param # Connected to
==================================================================================================
input_1 (InputLayer) [(None, 299, 299, 3) 0
__________________________________________________________________________________________________
conv2d (Conv2D) (None, 149, 149, 32) 864 input_1[0][0]
__________________________________________________________________________________________________
batch_normalization (BatchNorma (None, 149, 149, 32) 96 conv2d[0][0]
__________________________________________________________________________________________________
activation (Activation) (None, 149, 149, 32) 0 batch_normalization[0][0]
__________________________________________________________________________________________________
conv2d_1 (Conv2D) (None, 147, 147, 32) 9216 activation[0][0]
__________________________________________________________________________________________________
batch_normalization_1 (BatchNor (None, 147, 147, 32) 96 conv2d_1[0][0]
__________________________________________________________________________________________________
activation_1 (Activation) (None, 147, 147, 32) 0 batch_normalization_1[0][0]
__________________________________________________________________________________________________
conv2d_2 (Conv2D) (None, 147, 147, 64) 18432 activation_1[0][0]
__________________________________________________________________________________________________
batch_normalization_2 (BatchNor (None, 147, 147, 64) 192 conv2d_2[0][0]
__________________________________________________________________________________________________
activation_2 (Activation) (None, 147, 147, 64) 0 batch_normalization_2[0][0]
__________________________________________________________________________________________________
max_pooling2d (MaxPooling2D) (None, 73, 73, 64) 0 activation_2[0][0]
__________________________________________________________________________________________________
conv2d_3 (Conv2D) (None, 73, 73, 80) 5120 max_pooling2d[0][0]
__________________________________________________________________________________________________
batch_normalization_3 (BatchNor (None, 73, 73, 80) 240 conv2d_3[0][0]
__________________________________________________________________________________________________
activation_3 (Activation) (None, 73, 73, 80) 0 batch_normalization_3[0][0]
__________________________________________________________________________________________________
conv2d_4 (Conv2D) (None, 71, 71, 192) 138240 activation_3[0][0]
__________________________________________________________________________________________________
batch_normalization_4 (BatchNor (None, 71, 71, 192) 576 conv2d_4[0][0]
__________________________________________________________________________________________________
activation_4 (Activation) (None, 71, 71, 192) 0 batch_normalization_4[0][0]
__________________________________________________________________________________________________
max_pooling2d_1 (MaxPooling2D) (None, 35, 35, 192) 0 activation_4[0][0]
__________________________________________________________________________________________________
conv2d_8 (Conv2D) (None, 35, 35, 64) 12288 max_pooling2d_1[0][0]
__________________________________________________________________________________________________
batch_normalization_8 (BatchNor (None, 35, 35, 64) 192 conv2d_8[0][0]
__________________________________________________________________________________________________
activation_8 (Activation) (None, 35, 35, 64) 0 batch_normalization_8[0][0]
__________________________________________________________________________________________________
conv2d_6 (Conv2D) (None, 35, 35, 48) 9216 max_pooling2d_1[0][0]
__________________________________________________________________________________________________
conv2d_9 (Conv2D) (None, 35, 35, 96) 55296 activation_8[0][0]
__________________________________________________________________________________________________
batch_normalization_6 (BatchNor (None, 35, 35, 48) 144 conv2d_6[0][0]
__________________________________________________________________________________________________
batch_normalization_9 (BatchNor (None, 35, 35, 96) 288 conv2d_9[0][0]
__________________________________________________________________________________________________
activation_6 (Activation) (None, 35, 35, 48) 0 batch_normalization_6[0][0]
__________________________________________________________________________________________________
activation_9 (Activation) (None, 35, 35, 96) 0 batch_normalization_9[0][0]
__________________________________________________________________________________________________
average_pooling2d (AveragePooli (None, 35, 35, 192) 0 max_pooling2d_1[0][0]
__________________________________________________________________________________________________
conv2d_5 (Conv2D) (None, 35, 35, 64) 12288 max_pooling2d_1[0][0]
__________________________________________________________________________________________________
conv2d_7 (Conv2D) (None, 35, 35, 64) 76800 activation_6[0][0]
__________________________________________________________________________________________________
conv2d_10 (Conv2D) (None, 35, 35, 96) 82944 activation_9[0][0]
__________________________________________________________________________________________________
conv2d_11 (Conv2D) (None, 35, 35, 32) 6144 average_pooling2d[0][0]
__________________________________________________________________________________________________
batch_normalization_5 (BatchNor (None, 35, 35, 64) 192 conv2d_5[0][0]
__________________________________________________________________________________________________
batch_normalization_7 (BatchNor (None, 35, 35, 64) 192 conv2d_7[0][0]
__________________________________________________________________________________________________
batch_normalization_10 (BatchNo (None, 35, 35, 96) 288 conv2d_10[0][0]
__________________________________________________________________________________________________
batch_normalization_11 (BatchNo (None, 35, 35, 32) 96 conv2d_11[0][0]
__________________________________________________________________________________________________
activation_5 (Activation) (None, 35, 35, 64) 0 batch_normalization_5[0][0]
__________________________________________________________________________________________________
activation_7 (Activation) (None, 35, 35, 64) 0 batch_normalization_7[0][0]
__________________________________________________________________________________________________
activation_10 (Activation) (None, 35, 35, 96) 0 batch_normalization_10[0][0]
__________________________________________________________________________________________________
activation_11 (Activation) (None, 35, 35, 32) 0 batch_normalization_11[0][0]
__________________________________________________________________________________________________
mixed0 (Concatenate) (None, 35, 35, 256) 0 activation_5[0][0]
activation_7[0][0]
activation_10[0][0]
activation_11[0][0]
__________________________________________________________________________________________________
conv2d_15 (Conv2D) (None, 35, 35, 64) 16384 mixed0[0][0]
__________________________________________________________________________________________________
batch_normalization_15 (BatchNo (None, 35, 35, 64) 192 conv2d_15[0][0]
__________________________________________________________________________________________________
activation_15 (Activation) (None, 35, 35, 64) 0 batch_normalization_15[0][0]
__________________________________________________________________________________________________
conv2d_13 (Conv2D) (None, 35, 35, 48) 12288 mixed0[0][0]
__________________________________________________________________________________________________
conv2d_16 (Conv2D) (None, 35, 35, 96) 55296 activation_15[0][0]
__________________________________________________________________________________________________
batch_normalization_13 (BatchNo (None, 35, 35, 48) 144 conv2d_13[0][0]
__________________________________________________________________________________________________
batch_normalization_16 (BatchNo (None, 35, 35, 96) 288 conv2d_16[0][0]
__________________________________________________________________________________________________
activation_13 (Activation) (None, 35, 35, 48) 0 batch_normalization_13[0][0]
__________________________________________________________________________________________________
activation_16 (Activation) (None, 35, 35, 96) 0 batch_normalization_16[0][0]
__________________________________________________________________________________________________
average_pooling2d_1 (AveragePoo (None, 35, 35, 256) 0 mixed0[0][0]
__________________________________________________________________________________________________
conv2d_12 (Conv2D) (None, 35, 35, 64) 16384 mixed0[0][0]
__________________________________________________________________________________________________
conv2d_14 (Conv2D) (None, 35, 35, 64) 76800 activation_13[0][0]
__________________________________________________________________________________________________
conv2d_17 (Conv2D) (None, 35, 35, 96) 82944 activation_16[0][0]
__________________________________________________________________________________________________
conv2d_18 (Conv2D) (None, 35, 35, 64) 16384 average_pooling2d_1[0][0]
__________________________________________________________________________________________________
batch_normalization_12 (BatchNo (None, 35, 35, 64) 192 conv2d_12[0][0]
__________________________________________________________________________________________________
batch_normalization_14 (BatchNo (None, 35, 35, 64) 192 conv2d_14[0][0]
__________________________________________________________________________________________________
batch_normalization_17 (BatchNo (None, 35, 35, 96) 288 conv2d_17[0][0]
__________________________________________________________________________________________________
batch_normalization_18 (BatchNo (None, 35, 35, 64) 192 conv2d_18[0][0]
__________________________________________________________________________________________________
activation_12 (Activation) (None, 35, 35, 64) 0 batch_normalization_12[0][0]
__________________________________________________________________________________________________
activation_14 (Activation) (None, 35, 35, 64) 0 batch_normalization_14[0][0]
__________________________________________________________________________________________________
activation_17 (Activation) (None, 35, 35, 96) 0 batch_normalization_17[0][0]
__________________________________________________________________________________________________
activation_18 (Activation) (None, 35, 35, 64) 0 batch_normalization_18[0][0]
__________________________________________________________________________________________________
mixed1 (Concatenate) (None, 35, 35, 288) 0 activation_12[0][0]
activation_14[0][0]
activation_17[0][0]
activation_18[0][0]
__________________________________________________________________________________________________
conv2d_22 (Conv2D) (None, 35, 35, 64) 18432 mixed1[0][0]
__________________________________________________________________________________________________
batch_normalization_22 (BatchNo (None, 35, 35, 64) 192 conv2d_22[0][0]
__________________________________________________________________________________________________
activation_22 (Activation) (None, 35, 35, 64) 0 batch_normalization_22[0][0]
__________________________________________________________________________________________________
conv2d_20 (Conv2D) (None, 35, 35, 48) 13824 mixed1[0][0]
__________________________________________________________________________________________________
conv2d_23 (Conv2D) (None, 35, 35, 96) 55296 activation_22[0][0]
__________________________________________________________________________________________________
batch_normalization_20 (BatchNo (None, 35, 35, 48) 144 conv2d_20[0][0]
__________________________________________________________________________________________________
batch_normalization_23 (BatchNo (None, 35, 35, 96) 288 conv2d_23[0][0]
__________________________________________________________________________________________________
activation_20 (Activation) (None, 35, 35, 48) 0 batch_normalization_20[0][0]
__________________________________________________________________________________________________
activation_23 (Activation) (None, 35, 35, 96) 0 batch_normalization_23[0][0]
__________________________________________________________________________________________________
average_pooling2d_2 (AveragePoo (None, 35, 35, 288) 0 mixed1[0][0]
__________________________________________________________________________________________________
conv2d_19 (Conv2D) (None, 35, 35, 64) 18432 mixed1[0][0]
__________________________________________________________________________________________________
conv2d_21 (Conv2D) (None, 35, 35, 64) 76800 activation_20[0][0]
__________________________________________________________________________________________________
conv2d_24 (Conv2D) (None, 35, 35, 96) 82944 activation_23[0][0]
__________________________________________________________________________________________________
conv2d_25 (Conv2D) (None, 35, 35, 64) 18432 average_pooling2d_2[0][0]
__________________________________________________________________________________________________
batch_normalization_19 (BatchNo (None, 35, 35, 64) 192 conv2d_19[0][0]
__________________________________________________________________________________________________
batch_normalization_21 (BatchNo (None, 35, 35, 64) 192 conv2d_21[0][0]
__________________________________________________________________________________________________
batch_normalization_24 (BatchNo (None, 35, 35, 96) 288 conv2d_24[0][0]
__________________________________________________________________________________________________
batch_normalization_25 (BatchNo (None, 35, 35, 64) 192 conv2d_25[0][0]
__________________________________________________________________________________________________
activation_19 (Activation) (None, 35, 35, 64) 0 batch_normalization_19[0][0]
__________________________________________________________________________________________________
activation_21 (Activation) (None, 35, 35, 64) 0 batch_normalization_21[0][0]
__________________________________________________________________________________________________
activation_24 (Activation) (None, 35, 35, 96) 0 batch_normalization_24[0][0]
__________________________________________________________________________________________________
activation_25 (Activation) (None, 35, 35, 64) 0 batch_normalization_25[0][0]
__________________________________________________________________________________________________
mixed2 (Concatenate) (None, 35, 35, 288) 0 activation_19[0][0]
activation_21[0][0]
activation_24[0][0]
activation_25[0][0]
__________________________________________________________________________________________________
conv2d_27 (Conv2D) (None, 35, 35, 64) 18432 mixed2[0][0]
__________________________________________________________________________________________________
batch_normalization_27 (BatchNo (None, 35, 35, 64) 192 conv2d_27[0][0]
__________________________________________________________________________________________________
activation_27 (Activation) (None, 35, 35, 64) 0 batch_normalization_27[0][0]
__________________________________________________________________________________________________
conv2d_28 (Conv2D) (None, 35, 35, 96) 55296 activation_27[0][0]
__________________________________________________________________________________________________
batch_normalization_28 (BatchNo (None, 35, 35, 96) 288 conv2d_28[0][0]
__________________________________________________________________________________________________
activation_28 (Activation) (None, 35, 35, 96) 0 batch_normalization_28[0][0]
__________________________________________________________________________________________________
conv2d_26 (Conv2D) (None, 17, 17, 384) 995328 mixed2[0][0]
__________________________________________________________________________________________________
conv2d_29 (Conv2D) (None, 17, 17, 96) 82944 activation_28[0][0]
__________________________________________________________________________________________________
batch_normalization_26 (BatchNo (None, 17, 17, 384) 1152 conv2d_26[0][0]
__________________________________________________________________________________________________
batch_normalization_29 (BatchNo (None, 17, 17, 96) 288 conv2d_29[0][0]
__________________________________________________________________________________________________
activation_26 (Activation) (None, 17, 17, 384) 0 batch_normalization_26[0][0]
__________________________________________________________________________________________________
activation_29 (Activation) (None, 17, 17, 96) 0 batch_normalization_29[0][0]
__________________________________________________________________________________________________
max_pooling2d_2 (MaxPooling2D) (None, 17, 17, 288) 0 mixed2[0][0]
__________________________________________________________________________________________________
mixed3 (Concatenate) (None, 17, 17, 768) 0 activation_26[0][0]
activation_29[0][0]
max_pooling2d_2[0][0]
__________________________________________________________________________________________________
conv2d_34 (Conv2D) (None, 17, 17, 128) 98304 mixed3[0][0]
__________________________________________________________________________________________________
batch_normalization_34 (BatchNo (None, 17, 17, 128) 384 conv2d_34[0][0]
__________________________________________________________________________________________________
activation_34 (Activation) (None, 17, 17, 128) 0 batch_normalization_34[0][0]
__________________________________________________________________________________________________
conv2d_35 (Conv2D) (None, 17, 17, 128) 114688 activation_34[0][0]
__________________________________________________________________________________________________
batch_normalization_35 (BatchNo (None, 17, 17, 128) 384 conv2d_35[0][0]
__________________________________________________________________________________________________
activation_35 (Activation) (None, 17, 17, 128) 0 batch_normalization_35[0][0]
__________________________________________________________________________________________________
conv2d_31 (Conv2D) (None, 17, 17, 128) 98304 mixed3[0][0]
__________________________________________________________________________________________________
conv2d_36 (Conv2D) (None, 17, 17, 128) 114688 activation_35[0][0]
__________________________________________________________________________________________________
batch_normalization_31 (BatchNo (None, 17, 17, 128) 384 conv2d_31[0][0]
__________________________________________________________________________________________________
batch_normalization_36 (BatchNo (None, 17, 17, 128) 384 conv2d_36[0][0]
__________________________________________________________________________________________________
activation_31 (Activation) (None, 17, 17, 128) 0 batch_normalization_31[0][0]
__________________________________________________________________________________________________
activation_36 (Activation) (None, 17, 17, 128) 0 batch_normalization_36[0][0]
__________________________________________________________________________________________________
conv2d_32 (Conv2D) (None, 17, 17, 128) 114688 activation_31[0][0]
__________________________________________________________________________________________________
conv2d_37 (Conv2D) (None, 17, 17, 128) 114688 activation_36[0][0]
__________________________________________________________________________________________________
batch_normalization_32 (BatchNo (None, 17, 17, 128) 384 conv2d_32[0][0]
__________________________________________________________________________________________________
batch_normalization_37 (BatchNo (None, 17, 17, 128) 384 conv2d_37[0][0]
__________________________________________________________________________________________________
activation_32 (Activation) (None, 17, 17, 128) 0 batch_normalization_32[0][0]
__________________________________________________________________________________________________
activation_37 (Activation) (None, 17, 17, 128) 0 batch_normalization_37[0][0]
__________________________________________________________________________________________________
average_pooling2d_3 (AveragePoo (None, 17, 17, 768) 0 mixed3[0][0]
__________________________________________________________________________________________________
conv2d_30 (Conv2D) (None, 17, 17, 192) 147456 mixed3[0][0]
__________________________________________________________________________________________________
conv2d_33 (Conv2D) (None, 17, 17, 192) 172032 activation_32[0][0]
__________________________________________________________________________________________________
conv2d_38 (Conv2D) (None, 17, 17, 192) 172032 activation_37[0][0]
__________________________________________________________________________________________________
conv2d_39 (Conv2D) (None, 17, 17, 192) 147456 average_pooling2d_3[0][0]
__________________________________________________________________________________________________
batch_normalization_30 (BatchNo (None, 17, 17, 192) 576 conv2d_30[0][0]
__________________________________________________________________________________________________
batch_normalization_33 (BatchNo (None, 17, 17, 192) 576 conv2d_33[0][0]
__________________________________________________________________________________________________
batch_normalization_38 (BatchNo (None, 17, 17, 192) 576 conv2d_38[0][0]
__________________________________________________________________________________________________
batch_normalization_39 (BatchNo (None, 17, 17, 192) 576 conv2d_39[0][0]
__________________________________________________________________________________________________
activation_30 (Activation) (None, 17, 17, 192) 0 batch_normalization_30[0][0]
__________________________________________________________________________________________________
activation_33 (Activation) (None, 17, 17, 192) 0 batch_normalization_33[0][0]
__________________________________________________________________________________________________
activation_38 (Activation) (None, 17, 17, 192) 0 batch_normalization_38[0][0]
__________________________________________________________________________________________________
activation_39 (Activation) (None, 17, 17, 192) 0 batch_normalization_39[0][0]
__________________________________________________________________________________________________
mixed4 (Concatenate) (None, 17, 17, 768) 0 activation_30[0][0]
activation_33[0][0]
activation_38[0][0]
activation_39[0][0]
__________________________________________________________________________________________________
conv2d_44 (Conv2D) (None, 17, 17, 160) 122880 mixed4[0][0]
__________________________________________________________________________________________________
batch_normalization_44 (BatchNo (None, 17, 17, 160) 480 conv2d_44[0][0]
__________________________________________________________________________________________________
activation_44 (Activation) (None, 17, 17, 160) 0 batch_normalization_44[0][0]
__________________________________________________________________________________________________
conv2d_45 (Conv2D) (None, 17, 17, 160) 179200 activation_44[0][0]
__________________________________________________________________________________________________
batch_normalization_45 (BatchNo (None, 17, 17, 160) 480 conv2d_45[0][0]
__________________________________________________________________________________________________
activation_45 (Activation) (None, 17, 17, 160) 0 batch_normalization_45[0][0]
__________________________________________________________________________________________________
conv2d_41 (Conv2D) (None, 17, 17, 160) 122880 mixed4[0][0]
__________________________________________________________________________________________________
conv2d_46 (Conv2D) (None, 17, 17, 160) 179200 activation_45[0][0]
__________________________________________________________________________________________________
batch_normalization_41 (BatchNo (None, 17, 17, 160) 480 conv2d_41[0][0]
__________________________________________________________________________________________________
batch_normalization_46 (BatchNo (None, 17, 17, 160) 480 conv2d_46[0][0]
__________________________________________________________________________________________________
activation_41 (Activation) (None, 17, 17, 160) 0 batch_normalization_41[0][0]
__________________________________________________________________________________________________
activation_46 (Activation) (None, 17, 17, 160) 0 batch_normalization_46[0][0]
__________________________________________________________________________________________________
conv2d_42 (Conv2D) (None, 17, 17, 160) 179200 activation_41[0][0]
__________________________________________________________________________________________________
conv2d_47 (Conv2D) (None, 17, 17, 160) 179200 activation_46[0][0]
__________________________________________________________________________________________________
batch_normalization_42 (BatchNo (None, 17, 17, 160) 480 conv2d_42[0][0]
__________________________________________________________________________________________________
batch_normalization_47 (BatchNo (None, 17, 17, 160) 480 conv2d_47[0][0]
__________________________________________________________________________________________________
activation_42 (Activation) (None, 17, 17, 160) 0 batch_normalization_42[0][0]
__________________________________________________________________________________________________
activation_47 (Activation) (None, 17, 17, 160) 0 batch_normalization_47[0][0]
__________________________________________________________________________________________________
average_pooling2d_4 (AveragePoo (None, 17, 17, 768) 0 mixed4[0][0]
__________________________________________________________________________________________________
conv2d_40 (Conv2D) (None, 17, 17, 192) 147456 mixed4[0][0]
__________________________________________________________________________________________________
conv2d_43 (Conv2D) (None, 17, 17, 192) 215040 activation_42[0][0]
__________________________________________________________________________________________________
conv2d_48 (Conv2D) (None, 17, 17, 192) 215040 activation_47[0][0]
__________________________________________________________________________________________________
conv2d_49 (Conv2D) (None, 17, 17, 192) 147456 average_pooling2d_4[0][0]
__________________________________________________________________________________________________
batch_normalization_40 (BatchNo (None, 17, 17, 192) 576 conv2d_40[0][0]
__________________________________________________________________________________________________
batch_normalization_43 (BatchNo (None, 17, 17, 192) 576 conv2d_43[0][0]
__________________________________________________________________________________________________
batch_normalization_48 (BatchNo (None, 17, 17, 192) 576 conv2d_48[0][0]
__________________________________________________________________________________________________
batch_normalization_49 (BatchNo (None, 17, 17, 192) 576 conv2d_49[0][0]
__________________________________________________________________________________________________
activation_40 (Activation) (None, 17, 17, 192) 0 batch_normalization_40[0][0]
__________________________________________________________________________________________________
activation_43 (Activation) (None, 17, 17, 192) 0 batch_normalization_43[0][0]
__________________________________________________________________________________________________
activation_48 (Activation) (None, 17, 17, 192) 0 batch_normalization_48[0][0]
__________________________________________________________________________________________________
activation_49 (Activation) (None, 17, 17, 192) 0 batch_normalization_49[0][0]
__________________________________________________________________________________________________
mixed5 (Concatenate) (None, 17, 17, 768) 0 activation_40[0][0]
activation_43[0][0]
activation_48[0][0]
activation_49[0][0]
__________________________________________________________________________________________________
conv2d_54 (Conv2D) (None, 17, 17, 160) 122880 mixed5[0][0]
__________________________________________________________________________________________________
batch_normalization_54 (BatchNo (None, 17, 17, 160) 480 conv2d_54[0][0]
__________________________________________________________________________________________________
activation_54 (Activation) (None, 17, 17, 160) 0 batch_normalization_54[0][0]
__________________________________________________________________________________________________
conv2d_55 (Conv2D) (None, 17, 17, 160) 179200 activation_54[0][0]
__________________________________________________________________________________________________
batch_normalization_55 (BatchNo (None, 17, 17, 160) 480 conv2d_55[0][0]
__________________________________________________________________________________________________
activation_55 (Activation) (None, 17, 17, 160) 0 batch_normalization_55[0][0]
__________________________________________________________________________________________________
conv2d_51 (Conv2D) (None, 17, 17, 160) 122880 mixed5[0][0]
__________________________________________________________________________________________________
conv2d_56 (Conv2D) (None, 17, 17, 160) 179200 activation_55[0][0]
__________________________________________________________________________________________________
batch_normalization_51 (BatchNo (None, 17, 17, 160) 480 conv2d_51[0][0]
__________________________________________________________________________________________________
batch_normalization_56 (BatchNo (None, 17, 17, 160) 480 conv2d_56[0][0]
__________________________________________________________________________________________________
activation_51 (Activation) (None, 17, 17, 160) 0 batch_normalization_51[0][0]
__________________________________________________________________________________________________
activation_56 (Activation) (None, 17, 17, 160) 0 batch_normalization_56[0][0]
__________________________________________________________________________________________________
conv2d_52 (Conv2D) (None, 17, 17, 160) 179200 activation_51[0][0]
__________________________________________________________________________________________________
conv2d_57 (Conv2D) (None, 17, 17, 160) 179200 activation_56[0][0]
__________________________________________________________________________________________________
batch_normalization_52 (BatchNo (None, 17, 17, 160) 480 conv2d_52[0][0]
__________________________________________________________________________________________________
batch_normalization_57 (BatchNo (None, 17, 17, 160) 480 conv2d_57[0][0]
__________________________________________________________________________________________________
activation_52 (Activation) (None, 17, 17, 160) 0 batch_normalization_52[0][0]
__________________________________________________________________________________________________
activation_57 (Activation) (None, 17, 17, 160) 0 batch_normalization_57[0][0]
__________________________________________________________________________________________________
average_pooling2d_5 (AveragePoo (None, 17, 17, 768) 0 mixed5[0][0]
__________________________________________________________________________________________________
conv2d_50 (Conv2D) (None, 17, 17, 192) 147456 mixed5[0][0]
__________________________________________________________________________________________________
conv2d_53 (Conv2D) (None, 17, 17, 192) 215040 activation_52[0][0]
__________________________________________________________________________________________________
conv2d_58 (Conv2D) (None, 17, 17, 192) 215040 activation_57[0][0]
__________________________________________________________________________________________________
conv2d_59 (Conv2D) (None, 17, 17, 192) 147456 average_pooling2d_5[0][0]
__________________________________________________________________________________________________
batch_normalization_50 (BatchNo (None, 17, 17, 192) 576 conv2d_50[0][0]
__________________________________________________________________________________________________
batch_normalization_53 (BatchNo (None, 17, 17, 192) 576 conv2d_53[0][0]
__________________________________________________________________________________________________
batch_normalization_58 (BatchNo (None, 17, 17, 192) 576 conv2d_58[0][0]
__________________________________________________________________________________________________
batch_normalization_59 (BatchNo (None, 17, 17, 192) 576 conv2d_59[0][0]
__________________________________________________________________________________________________
activation_50 (Activation) (None, 17, 17, 192) 0 batch_normalization_50[0][0]
__________________________________________________________________________________________________
activation_53 (Activation) (None, 17, 17, 192) 0 batch_normalization_53[0][0]
__________________________________________________________________________________________________
activation_58 (Activation) (None, 17, 17, 192) 0 batch_normalization_58[0][0]
__________________________________________________________________________________________________
activation_59 (Activation) (None, 17, 17, 192) 0 batch_normalization_59[0][0]
__________________________________________________________________________________________________
mixed6 (Concatenate) (None, 17, 17, 768) 0 activation_50[0][0]
activation_53[0][0]
activation_58[0][0]
activation_59[0][0]
__________________________________________________________________________________________________
conv2d_64 (Conv2D) (None, 17, 17, 192) 147456 mixed6[0][0]
__________________________________________________________________________________________________
batch_normalization_64 (BatchNo (None, 17, 17, 192) 576 conv2d_64[0][0]
__________________________________________________________________________________________________
activation_64 (Activation) (None, 17, 17, 192) 0 batch_normalization_64[0][0]
__________________________________________________________________________________________________
conv2d_65 (Conv2D) (None, 17, 17, 192) 258048 activation_64[0][0]
__________________________________________________________________________________________________
batch_normalization_65 (BatchNo (None, 17, 17, 192) 576 conv2d_65[0][0]
__________________________________________________________________________________________________
activation_65 (Activation) (None, 17, 17, 192) 0 batch_normalization_65[0][0]
__________________________________________________________________________________________________
conv2d_61 (Conv2D) (None, 17, 17, 192) 147456 mixed6[0][0]
__________________________________________________________________________________________________
conv2d_66 (Conv2D) (None, 17, 17, 192) 258048 activation_65[0][0]
__________________________________________________________________________________________________
batch_normalization_61 (BatchNo (None, 17, 17, 192) 576 conv2d_61[0][0]
__________________________________________________________________________________________________
batch_normalization_66 (BatchNo (None, 17, 17, 192) 576 conv2d_66[0][0]
__________________________________________________________________________________________________
activation_61 (Activation) (None, 17, 17, 192) 0 batch_normalization_61[0][0]
__________________________________________________________________________________________________
activation_66 (Activation) (None, 17, 17, 192) 0 batch_normalization_66[0][0]
__________________________________________________________________________________________________
conv2d_62 (Conv2D) (None, 17, 17, 192) 258048 activation_61[0][0]
__________________________________________________________________________________________________
conv2d_67 (Conv2D) (None, 17, 17, 192) 258048 activation_66[0][0]
__________________________________________________________________________________________________
batch_normalization_62 (BatchNo (None, 17, 17, 192) 576 conv2d_62[0][0]
__________________________________________________________________________________________________
batch_normalization_67 (BatchNo (None, 17, 17, 192) 576 conv2d_67[0][0]
__________________________________________________________________________________________________
activation_62 (Activation) (None, 17, 17, 192) 0 batch_normalization_62[0][0]
__________________________________________________________________________________________________
activation_67 (Activation) (None, 17, 17, 192) 0 batch_normalization_67[0][0]
__________________________________________________________________________________________________
average_pooling2d_6 (AveragePoo (None, 17, 17, 768) 0 mixed6[0][0]
__________________________________________________________________________________________________
conv2d_60 (Conv2D) (None, 17, 17, 192) 147456 mixed6[0][0]
__________________________________________________________________________________________________
conv2d_63 (Conv2D) (None, 17, 17, 192) 258048 activation_62[0][0]
__________________________________________________________________________________________________
conv2d_68 (Conv2D) (None, 17, 17, 192) 258048 activation_67[0][0]
__________________________________________________________________________________________________
conv2d_69 (Conv2D) (None, 17, 17, 192) 147456 average_pooling2d_6[0][0]
__________________________________________________________________________________________________
batch_normalization_60 (BatchNo (None, 17, 17, 192) 576 conv2d_60[0][0]
__________________________________________________________________________________________________
batch_normalization_63 (BatchNo (None, 17, 17, 192) 576 conv2d_63[0][0]
__________________________________________________________________________________________________
batch_normalization_68 (BatchNo (None, 17, 17, 192) 576 conv2d_68[0][0]
__________________________________________________________________________________________________
batch_normalization_69 (BatchNo (None, 17, 17, 192) 576 conv2d_69[0][0]
__________________________________________________________________________________________________
activation_60 (Activation) (None, 17, 17, 192) 0 batch_normalization_60[0][0]
__________________________________________________________________________________________________
activation_63 (Activation) (None, 17, 17, 192) 0 batch_normalization_63[0][0]
__________________________________________________________________________________________________
activation_68 (Activation) (None, 17, 17, 192) 0 batch_normalization_68[0][0]
__________________________________________________________________________________________________
activation_69 (Activation) (None, 17, 17, 192) 0 batch_normalization_69[0][0]
__________________________________________________________________________________________________
mixed7 (Concatenate) (None, 17, 17, 768) 0 activation_60[0][0]
activation_63[0][0]
activation_68[0][0]
activation_69[0][0]
__________________________________________________________________________________________________
conv2d_72 (Conv2D) (None, 17, 17, 192) 147456 mixed7[0][0]
__________________________________________________________________________________________________
batch_normalization_72 (BatchNo (None, 17, 17, 192) 576 conv2d_72[0][0]
__________________________________________________________________________________________________
activation_72 (Activation) (None, 17, 17, 192) 0 batch_normalization_72[0][0]
__________________________________________________________________________________________________
conv2d_73 (Conv2D) (None, 17, 17, 192) 258048 activation_72[0][0]
__________________________________________________________________________________________________
batch_normalization_73 (BatchNo (None, 17, 17, 192) 576 conv2d_73[0][0]
__________________________________________________________________________________________________
activation_73 (Activation) (None, 17, 17, 192) 0 batch_normalization_73[0][0]
__________________________________________________________________________________________________
conv2d_70 (Conv2D) (None, 17, 17, 192) 147456 mixed7[0][0]
__________________________________________________________________________________________________
conv2d_74 (Conv2D) (None, 17, 17, 192) 258048 activation_73[0][0]
__________________________________________________________________________________________________
batch_normalization_70 (BatchNo (None, 17, 17, 192) 576 conv2d_70[0][0]
__________________________________________________________________________________________________
batch_normalization_74 (BatchNo (None, 17, 17, 192) 576 conv2d_74[0][0]
__________________________________________________________________________________________________
activation_70 (Activation) (None, 17, 17, 192) 0 batch_normalization_70[0][0]
__________________________________________________________________________________________________
activation_74 (Activation) (None, 17, 17, 192) 0 batch_normalization_74[0][0]
__________________________________________________________________________________________________
conv2d_71 (Conv2D) (None, 8, 8, 320) 552960 activation_70[0][0]
__________________________________________________________________________________________________
conv2d_75 (Conv2D) (None, 8, 8, 192) 331776 activation_74[0][0]
__________________________________________________________________________________________________
batch_normalization_71 (BatchNo (None, 8, 8, 320) 960 conv2d_71[0][0]
__________________________________________________________________________________________________
batch_normalization_75 (BatchNo (None, 8, 8, 192) 576 conv2d_75[0][0]
__________________________________________________________________________________________________
activation_71 (Activation) (None, 8, 8, 320) 0 batch_normalization_71[0][0]
__________________________________________________________________________________________________
activation_75 (Activation) (None, 8, 8, 192) 0 batch_normalization_75[0][0]
__________________________________________________________________________________________________
max_pooling2d_3 (MaxPooling2D) (None, 8, 8, 768) 0 mixed7[0][0]
__________________________________________________________________________________________________
mixed8 (Concatenate) (None, 8, 8, 1280) 0 activation_71[0][0]
activation_75[0][0]
max_pooling2d_3[0][0]
__________________________________________________________________________________________________
conv2d_80 (Conv2D) (None, 8, 8, 448) 573440 mixed8[0][0]
__________________________________________________________________________________________________
batch_normalization_80 (BatchNo (None, 8, 8, 448) 1344 conv2d_80[0][0]
__________________________________________________________________________________________________
activation_80 (Activation) (None, 8, 8, 448) 0 batch_normalization_80[0][0]
__________________________________________________________________________________________________
conv2d_77 (Conv2D) (None, 8, 8, 384) 491520 mixed8[0][0]
__________________________________________________________________________________________________
conv2d_81 (Conv2D) (None, 8, 8, 384) 1548288 activation_80[0][0]
__________________________________________________________________________________________________
batch_normalization_77 (BatchNo (None, 8, 8, 384) 1152 conv2d_77[0][0]
__________________________________________________________________________________________________
batch_normalization_81 (BatchNo (None, 8, 8, 384) 1152 conv2d_81[0][0]
__________________________________________________________________________________________________
activation_77 (Activation) (None, 8, 8, 384) 0 batch_normalization_77[0][0]
__________________________________________________________________________________________________
activation_81 (Activation) (None, 8, 8, 384) 0 batch_normalization_81[0][0]
__________________________________________________________________________________________________
conv2d_78 (Conv2D) (None, 8, 8, 384) 442368 activation_77[0][0]
__________________________________________________________________________________________________
conv2d_79 (Conv2D) (None, 8, 8, 384) 442368 activation_77[0][0]
__________________________________________________________________________________________________
conv2d_82 (Conv2D) (None, 8, 8, 384) 442368 activation_81[0][0]
__________________________________________________________________________________________________
conv2d_83 (Conv2D) (None, 8, 8, 384) 442368 activation_81[0][0]
__________________________________________________________________________________________________
average_pooling2d_7 (AveragePoo (None, 8, 8, 1280) 0 mixed8[0][0]
__________________________________________________________________________________________________
conv2d_76 (Conv2D) (None, 8, 8, 320) 409600 mixed8[0][0]
__________________________________________________________________________________________________
batch_normalization_78 (BatchNo (None, 8, 8, 384) 1152 conv2d_78[0][0]
__________________________________________________________________________________________________
batch_normalization_79 (BatchNo (None, 8, 8, 384) 1152 conv2d_79[0][0]
__________________________________________________________________________________________________
batch_normalization_82 (BatchNo (None, 8, 8, 384) 1152 conv2d_82[0][0]
__________________________________________________________________________________________________
batch_normalization_83 (BatchNo (None, 8, 8, 384) 1152 conv2d_83[0][0]
__________________________________________________________________________________________________
conv2d_84 (Conv2D) (None, 8, 8, 192) 245760 average_pooling2d_7[0][0]
__________________________________________________________________________________________________
batch_normalization_76 (BatchNo (None, 8, 8, 320) 960 conv2d_76[0][0]
__________________________________________________________________________________________________
activation_78 (Activation) (None, 8, 8, 384) 0 batch_normalization_78[0][0]
__________________________________________________________________________________________________
activation_79 (Activation) (None, 8, 8, 384) 0 batch_normalization_79[0][0]
__________________________________________________________________________________________________
activation_82 (Activation) (None, 8, 8, 384) 0 batch_normalization_82[0][0]
__________________________________________________________________________________________________
activation_83 (Activation) (None, 8, 8, 384) 0 batch_normalization_83[0][0]
__________________________________________________________________________________________________
batch_normalization_84 (BatchNo (None, 8, 8, 192) 576 conv2d_84[0][0]
__________________________________________________________________________________________________
activation_76 (Activation) (None, 8, 8, 320) 0 batch_normalization_76[0][0]
__________________________________________________________________________________________________
mixed9_0 (Concatenate) (None, 8, 8, 768) 0 activation_78[0][0]
activation_79[0][0]
__________________________________________________________________________________________________
concatenate (Concatenate) (None, 8, 8, 768) 0 activation_82[0][0]
activation_83[0][0]
__________________________________________________________________________________________________
activation_84 (Activation) (None, 8, 8, 192) 0 batch_normalization_84[0][0]
__________________________________________________________________________________________________
mixed9 (Concatenate) (None, 8, 8, 2048) 0 activation_76[0][0]
mixed9_0[0][0]
concatenate[0][0]
activation_84[0][0]
__________________________________________________________________________________________________
conv2d_89 (Conv2D) (None, 8, 8, 448) 917504 mixed9[0][0]
__________________________________________________________________________________________________
batch_normalization_89 (BatchNo (None, 8, 8, 448) 1344 conv2d_89[0][0]
__________________________________________________________________________________________________
activation_89 (Activation) (None, 8, 8, 448) 0 batch_normalization_89[0][0]
__________________________________________________________________________________________________
conv2d_86 (Conv2D) (None, 8, 8, 384) 786432 mixed9[0][0]
__________________________________________________________________________________________________
conv2d_90 (Conv2D) (None, 8, 8, 384) 1548288 activation_89[0][0]
__________________________________________________________________________________________________
batch_normalization_86 (BatchNo (None, 8, 8, 384) 1152 conv2d_86[0][0]
__________________________________________________________________________________________________
batch_normalization_90 (BatchNo (None, 8, 8, 384) 1152 conv2d_90[0][0]
__________________________________________________________________________________________________
activation_86 (Activation) (None, 8, 8, 384) 0 batch_normalization_86[0][0]
__________________________________________________________________________________________________
activation_90 (Activation) (None, 8, 8, 384) 0 batch_normalization_90[0][0]
__________________________________________________________________________________________________
conv2d_87 (Conv2D) (None, 8, 8, 384) 442368 activation_86[0][0]
__________________________________________________________________________________________________
conv2d_88 (Conv2D) (None, 8, 8, 384) 442368 activation_86[0][0]
__________________________________________________________________________________________________
conv2d_91 (Conv2D) (None, 8, 8, 384) 442368 activation_90[0][0]
__________________________________________________________________________________________________
conv2d_92 (Conv2D) (None, 8, 8, 384) 442368 activation_90[0][0]
__________________________________________________________________________________________________
average_pooling2d_8 (AveragePoo (None, 8, 8, 2048) 0 mixed9[0][0]
__________________________________________________________________________________________________
conv2d_85 (Conv2D) (None, 8, 8, 320) 655360 mixed9[0][0]
__________________________________________________________________________________________________
batch_normalization_87 (BatchNo (None, 8, 8, 384) 1152 conv2d_87[0][0]
__________________________________________________________________________________________________
batch_normalization_88 (BatchNo (None, 8, 8, 384) 1152 conv2d_88[0][0]
__________________________________________________________________________________________________
batch_normalization_91 (BatchNo (None, 8, 8, 384) 1152 conv2d_91[0][0]
__________________________________________________________________________________________________
batch_normalization_92 (BatchNo (None, 8, 8, 384) 1152 conv2d_92[0][0]
__________________________________________________________________________________________________
conv2d_93 (Conv2D) (None, 8, 8, 192) 393216 average_pooling2d_8[0][0]
__________________________________________________________________________________________________
batch_normalization_85 (BatchNo (None, 8, 8, 320) 960 conv2d_85[0][0]
__________________________________________________________________________________________________
activation_87 (Activation) (None, 8, 8, 384) 0 batch_normalization_87[0][0]
__________________________________________________________________________________________________
activation_88 (Activation) (None, 8, 8, 384) 0 batch_normalization_88[0][0]
__________________________________________________________________________________________________
activation_91 (Activation) (None, 8, 8, 384) 0 batch_normalization_91[0][0]
__________________________________________________________________________________________________
activation_92 (Activation) (None, 8, 8, 384) 0 batch_normalization_92[0][0]
__________________________________________________________________________________________________
batch_normalization_93 (BatchNo (None, 8, 8, 192) 576 conv2d_93[0][0]
__________________________________________________________________________________________________
activation_85 (Activation) (None, 8, 8, 320) 0 batch_normalization_85[0][0]
__________________________________________________________________________________________________
mixed9_1 (Concatenate) (None, 8, 8, 768) 0 activation_87[0][0]
activation_88[0][0]
__________________________________________________________________________________________________
concatenate_1 (Concatenate) (None, 8, 8, 768) 0 activation_91[0][0]
activation_92[0][0]
__________________________________________________________________________________________________
activation_93 (Activation) (None, 8, 8, 192) 0 batch_normalization_93[0][0]
__________________________________________________________________________________________________
mixed10 (Concatenate) (None, 8, 8, 2048) 0 activation_85[0][0]
mixed9_1[0][0]
concatenate_1[0][0]
activation_93[0][0]
__________________________________________________________________________________________________
avg_pool (GlobalAveragePooling2 (None, 2048) 0 mixed10[0][0]
==================================================================================================
Total params: 21,802,784
Trainable params: 21,768,352
Non-trainable params: 34,432
__________________________________________________________________________________________________
None
# plot model
filename = 'models/cnn_inceptionv3.png'
plot(cnn_inceptionv3, filename)
Output hidden; open in https://colab.research.google.com to view.
filename = 'files/features_inceptionv3.pkl'
# only extract if file does not exist
if not isfile(filename):
# extract features from all images
directory = 'Flickr8k_Dataset'
features = extract_features(directory, cnn_vgg16, 224)
# save to file
dump(features, open(filename, 'wb'))
# display original and preprocessed image
example_image = "Flickr8k_Dataset/667626_18933d713e.jpg"
display(Image(example_image))
image = preprocess_image(example_image, 299)
plt.imshow(np.squeeze(image))
4. Prepare Text Data
# load doc into memory
def load_doc(filename):
# open the file as read only
file = open(filename, 'r')
# read all text
text = file.read()
# close the file
file.close()
return text
# extract descriptions for images
def load_descriptions(doc):
mapping = dict()
# process lines
for line in doc.split('\n'):
# split line by white space
tokens = line.split()
if len(line) < 2:
continue
# take the first token as the image id, the rest as the description
image_id, image_desc = tokens[0], tokens[1:]
# remove filename from image id
image_id = image_id.split('.')[0]
# convert description tokens back to string
image_desc = ' '.join(image_desc)
# create the list if needed
if image_id not in mapping:
mapping[image_id] = list()
# store description
mapping[image_id].append(image_desc)
return mapping
def clean_descriptions(descriptions):
# prepare translation table for removing punctuation
table = str.maketrans('', '', string.punctuation)
for key, desc_list in descriptions.items():
for i in range(len(desc_list)):
desc = desc_list[i]
# tokenize
desc = desc.split()
# convert to lower case
desc = [word.lower() for word in desc]
# remove punctuation from each token
desc = [w.translate(table) for w in desc]
# remove hanging 's' and 'a'
desc = [word for word in desc if len(word)>1]
# remove tokens with numbers in them
desc = [word for word in desc if word.isalpha()]
# store as string
desc_list[i] = ' '.join(desc)
# convert the loaded descriptions into a vocabulary of words
def to_vocabulary(descriptions):
# build a list of all description strings
all_desc = set()
for key in descriptions.keys():
[all_desc.update(d.split()) for d in descriptions[key]]
return all_desc
# save descriptions to file, one per line
def save_descriptions(descriptions, filename):
lines = list()
for key, desc_list in descriptions.items():
for desc in desc_list:
lines.append(key + ' ' + desc)
data = '\n'.join(lines)
file = open(filename, 'w')
file.write(data)
file.close()
filename = 'Flickr8k_text/Flickr8k.token.txt'
# load descriptions
doc = load_doc(filename)
# parse descriptions
descriptions = load_descriptions(doc)
print('Loaded: %d ' % len(descriptions))
Loaded: 8092
display(Image(example_image))
example_id = '667626_18933d713e'
descriptions[example_id]
['A girl is stretched out in shallow water',
'A girl wearing a red and multi-colored bikini is laying on her back in shallow water .',
'A little girl in a red swimsuit is laying on her back in shallow water .',
'A young girl is lying in the sand , while ocean water is surrounding her .',
'Girl wearing a bikini lying on her back in a shallow pool of clear blue water .']
# clean descriptions
clean_descriptions(descriptions)
# summarize vocabulary
vocabulary = to_vocabulary(descriptions)
print('Vocabulary Size: %d' % len(vocabulary))
# save to file
save_descriptions(descriptions, 'files/descriptions.txt')
Vocabulary Size: 8763
descriptions[example_id]
['girl is stretched out in shallow water',
'girl wearing red and multicolored bikini is laying on her back in shallow water',
'little girl in red swimsuit is laying on her back in shallow water',
'young girl is lying in the sand while ocean water is surrounding her',
'girl wearing bikini lying on her back in shallow pool of clear blue water']
5. Load Data
# load a pre-defined list of photo identifiers
def load_set(filename):
doc = load_doc(filename)
dataset = list()
# process line by line
for line in doc.split('\n'):
# skip empty lines
if len(line) < 1:
continue
# get the image identifier
identifier = line.split('.')[0]
dataset.append(identifier)
return set(dataset)
# load clean descriptions into memory
def load_clean_descriptions(filename, dataset):
# load document
doc = load_doc(filename)
descriptions = dict()
for line in doc.split('\n'):
# split line by white space
tokens = line.split()
# split id from description
image_id, image_desc = tokens[0], tokens[1:]
# skip images not in the set
if image_id in dataset:
# create list
if image_id not in descriptions:
descriptions[image_id] = list()
# wrap description in tokens
desc = '<start> ' + ' '.join(image_desc) + ' <end>'
# store
descriptions[image_id].append(desc)
return descriptions
# load photo features
def load_photo_features(filename, dataset):
# load all features
all_features = load(open(filename, 'rb'))
# filter features
features = {k: all_features[k] for k in dataset}
return features
5.1. Train Data
# load training dataset (6K)
filename = 'Flickr8k_text/Flickr_8k.trainImages.txt'
train = load_set(filename)
print('Dataset: %d' % len(train))
# descriptions
train_descriptions = load_clean_descriptions('files/descriptions.txt', train)
print('Descriptions: train=%d' % len(train_descriptions))
# photo features
train_features = load_photo_features('files/features_vgg16.pkl', train) #files/features_inceptionv3 or files/features_vgg16
print('Photos: train=%d' % len(train_features))
Dataset: 6000
Descriptions: train=6000
Photos: train=6000
display(Image(example_image))
# <start> and <end> added
train_descriptions[example_id]
['<start> girl is stretched out in shallow water <end>',
'<start> girl wearing red and multicolored bikini is laying on her back in shallow water <end>',
'<start> little girl in red swimsuit is laying on her back in shallow water <end>',
'<start> young girl is lying in the sand while ocean water is surrounding her <end>',
'<start> girl wearing bikini lying on her back in shallow pool of clear blue water <end>']
# features of previous image
train_features[example_id]
array([[0.39258254, 0.17952543, 0.5020331 , ..., 0.31590924, 0.616592 ,
0.37674323]], dtype=float32)
5.2. Validation Data
# load val dataset
filename = 'Flickr8k_text/Flickr_8k.devImages.txt'
val = load_set(filename)
print('Dataset: %d' % len(val))
# descriptions
val_descriptions = load_clean_descriptions('files/descriptions.txt', val)
print('Descriptions: val=%d' % len(val_descriptions))
# photo features
val_features = load_photo_features('files/features_vgg16.pkl', val)#files/features_inceptionv3 or files/features_vgg16
print('Photos: val=%d' % len(val_features))
Dataset: 1000
Descriptions: val=1000
Photos: val=1000
5.3. Test Data
# load test set
filename = 'Flickr8k_text/Flickr_8k.testImages.txt'
test = load_set(filename)
print('Dataset: %d' % len(test))
# descriptions
test_descriptions = load_clean_descriptions('files/descriptions.txt', test)
print('Descriptions: test=%d' % len(test_descriptions))
# photo features
test_features = load_photo_features('files/features_vgg16.pkl', test)#files/features_inceptionv3 or files/features_vgg16
print('Photos: test=%d' % len(test_features))
Dataset: 1000
Descriptions: test=1000
Photos: test=1000
6. Encode Text Data
# covert a dictionary of clean descriptions to a list of descriptions
def to_lines(descriptions):
all_desc = list()
for key in descriptions.keys():
[all_desc.append(d) for d in descriptions[key]]
return all_desc
# fit a tokenizer given caption descriptions
def create_tokenizer(descriptions):
lines = to_lines(descriptions)
tokenizer = Tokenizer(filters='!"#$%&()*+.,-/:;=?@[\]^_`{|}~ ')
tokenizer.fit_on_texts(lines)
return tokenizer
# calculate the length of the description with the most words
def max_len(descriptions):
lines = to_lines(descriptions)
return max(len(d.split()) for d in lines)
# prepare tokenizer
filename = 'files/tokenizer.pkl'
# only create tokenizer if it does not exist
if not isfile(filename):
tokenizer = create_tokenizer(train_descriptions)
# save the tokenizer
dump(tokenizer, open(filename, 'wb'))
else:
tokenizer = load(open(filename, 'rb'))
# define vocabulary size
vocab_size = len(tokenizer.word_index) + 1
print('Vocabulary Size: %d' % vocab_size)
# determine the maximum sequence length
max_length = max_len(train_descriptions)
print('Description Length: %d' % max_length)
Vocabulary Size: 7579
Description Length: 34
# 10 most common words
list(tokenizer.word_index.items())[:10]
[('<start>', 1),
('<end>', 2),
('in', 3),
('the', 4),
('on', 5),
('is', 6),
('and', 7),
('dog', 8),
('with', 9),
('man', 10)]
# 10 least common words
list(tokenizer.word_index.items())[-10:]
[('exotic', 7569),
('swatting', 7570),
('containig', 7571),
('rainstorm', 7572),
('breezeway', 7573),
('cocker', 7574),
('spaniels', 7575),
('majestically', 7576),
('scrolled', 7577),
('patterns', 7578)]
7. Define Model
from keras.backend import clear_session
clear_session()
7.1. Model 1
# define the captioning model
def rnn_model_1(vocab_size, max_length, embedding_size, units, input_size):
# feature extractor model
inputs1 = Input(shape=(input_size,))
fe1 = Dropout(0.5)(inputs1)
fe2 = Dense(embedding_size, activation='relu')(fe1)
# sequence model
inputs2 = Input(shape=(max_length,))
se1 = Embedding(vocab_size, embedding_size, mask_zero=True)(inputs2)
se2 = Dropout(0.5)(se1)
se3 = LSTM(units)(se2)
# decoder model
decoder1 = add([fe2, se3])
decoder2 = Dense(units, activation='relu')(decoder1)
outputs = Dense(vocab_size, activation='softmax')(decoder2)
# tie it together [image, seq] [word]
model = Model(inputs=[inputs1, inputs2], outputs=outputs)
model.compile(loss='categorical_crossentropy', optimizer='adam')
# summarize model
print(model.summary())
return model
7.1.1. Model 1 VGG16
# define the model
embedding_size = 256
units = 256
input_size = 4096
model = rnn_model_1(vocab_size, max_length, embedding_size, units, input_size)
Model: "model"
__________________________________________________________________________________________________
Layer (type) Output Shape Param # Connected to
==================================================================================================
input_2 (InputLayer) [(None, 34)] 0
__________________________________________________________________________________________________
input_1 (InputLayer) [(None, 4096)] 0
__________________________________________________________________________________________________
embedding (Embedding) (None, 34, 256) 1940224 input_2[0][0]
__________________________________________________________________________________________________
dropout (Dropout) (None, 4096) 0 input_1[0][0]
__________________________________________________________________________________________________
dropout_1 (Dropout) (None, 34, 256) 0 embedding[0][0]
__________________________________________________________________________________________________
dense (Dense) (None, 256) 1048832 dropout[0][0]
__________________________________________________________________________________________________
lstm (LSTM) (None, 256) 525312 dropout_1[0][0]
__________________________________________________________________________________________________
add (Add) (None, 256) 0 dense[0][0]
lstm[0][0]
__________________________________________________________________________________________________
dense_1 (Dense) (None, 256) 65792 add[0][0]
__________________________________________________________________________________________________
dense_2 (Dense) (None, 7579) 1947803 dense_1[0][0]
==================================================================================================
Total params: 5,527,963
Trainable params: 5,527,963
Non-trainable params: 0
__________________________________________________________________________________________________
None
# plot model
filename = 'models/rnn_model_1_vgg16.png'
plot(model, filename)
7.1.2. Model 1 InceptionV3
# define the model
embedding_size = 256
units = 256
input_size = 2048
model = rnn_model_1(vocab_size, max_length, embedding_size, units, input_size)
Model: "model_1"
__________________________________________________________________________________________________
Layer (type) Output Shape Param # Connected to
==================================================================================================
input_4 (InputLayer) [(None, 34)] 0
__________________________________________________________________________________________________
input_3 (InputLayer) [(None, 2048)] 0
__________________________________________________________________________________________________
embedding_1 (Embedding) (None, 34, 256) 1940224 input_4[0][0]
__________________________________________________________________________________________________
dropout_2 (Dropout) (None, 2048) 0 input_3[0][0]
__________________________________________________________________________________________________
dropout_3 (Dropout) (None, 34, 256) 0 embedding_1[0][0]
__________________________________________________________________________________________________
dense_3 (Dense) (None, 256) 524544 dropout_2[0][0]
__________________________________________________________________________________________________
lstm_1 (LSTM) (None, 256) 525312 dropout_3[0][0]
__________________________________________________________________________________________________
add_1 (Add) (None, 256) 0 dense_3[0][0]
lstm_1[0][0]
__________________________________________________________________________________________________
dense_4 (Dense) (None, 256) 65792 add_1[0][0]
__________________________________________________________________________________________________
dense_5 (Dense) (None, 7579) 1947803 dense_4[0][0]
==================================================================================================
Total params: 5,003,675
Trainable params: 5,003,675
Non-trainable params: 0
__________________________________________________________________________________________________
None
# plot model
filename = 'models/rnn_model_1_inceptionv3.png'
plot(model, filename)
7.2. Model 2
# define the captioning model
def rnn_model_2(vocab_size, max_length, embedding_size, units, input_size):
image_input = Input(shape=(input_size,))
image_model_1 = Dense(embedding_size, activation='relu')(image_input)
image_model = RepeatVector(max_length)(image_model_1)
caption_input = Input(shape=(max_length,))
# mask_zero: We zero pad inputs to the same length, the zero mask ignores those inputs
caption_model_1 = Embedding(vocab_size, embedding_size, mask_zero=True)(caption_input)
# Since we are going to predict the next word using the previous words, we have to set return_sequences = True.
caption_model_2 = LSTM(units, return_sequences=True)(caption_model_1)
caption_model = TimeDistributed(Dense(embedding_size))(caption_model_2)
# Merging the models and creating a softmax classifier
final_model_1 = concatenate([image_model, caption_model])
final_model_2 = Bidirectional(LSTM(units, return_sequences=False))(final_model_1)
final_model_3 = Dense(units, activation='relu')(final_model_2)
final_model = Dense(vocab_size, activation='softmax')(final_model_3)
model = Model(inputs=[image_input, caption_input], outputs=final_model)
model.compile(loss='categorical_crossentropy', optimizer='adam')
# summarize model
print(model.summary())
return model
7.2.1. Model 2 VGG16
# define the model
embedding_size = 256
units = 256
input_size = 4096
model = rnn_model_2(vocab_size, max_length, embedding_size, units, input_size)
Model: "model_1"
__________________________________________________________________________________________________
Layer (type) Output Shape Param # Connected to
==================================================================================================
input_3 (InputLayer) [(None, 34)] 0
__________________________________________________________________________________________________
input_2 (InputLayer) [(None, 4096)] 0
__________________________________________________________________________________________________
embedding (Embedding) (None, 34, 256) 1940224 input_3[0][0]
__________________________________________________________________________________________________
dense (Dense) (None, 256) 1048832 input_2[0][0]
__________________________________________________________________________________________________
lstm (LSTM) (None, 34, 256) 525312 embedding[0][0]
__________________________________________________________________________________________________
repeat_vector (RepeatVector) (None, 34, 256) 0 dense[0][0]
__________________________________________________________________________________________________
time_distributed (TimeDistribut (None, 34, 256) 65792 lstm[0][0]
__________________________________________________________________________________________________
concatenate (Concatenate) (None, 34, 512) 0 repeat_vector[0][0]
time_distributed[0][0]
__________________________________________________________________________________________________
bidirectional (Bidirectional) (None, 512) 1574912 concatenate[0][0]
__________________________________________________________________________________________________
dense_2 (Dense) (None, 256) 131328 bidirectional[0][0]
__________________________________________________________________________________________________
dense_3 (Dense) (None, 7579) 1947803 dense_2[0][0]
==================================================================================================
Total params: 7,234,203
Trainable params: 7,234,203
Non-trainable params: 0
__________________________________________________________________________________________________
None
# plot model
filename = 'models/rnn_model_2_vgg16.png'
plot(model, filename)
7.2.2. Model 2 InceptionV3
# define the model
embedding_size = 256
units = 256
input_size = 2048
model = rnn_model_2(vocab_size, max_length, embedding_size, units, input_size)
Model: "model_2"
__________________________________________________________________________________________________
Layer (type) Output Shape Param # Connected to
==================================================================================================
input_5 (InputLayer) [(None, 34)] 0
__________________________________________________________________________________________________
input_4 (InputLayer) [(None, 2048)] 0
__________________________________________________________________________________________________
embedding_1 (Embedding) (None, 34, 256) 1940224 input_5[0][0]
__________________________________________________________________________________________________
dense_4 (Dense) (None, 256) 524544 input_4[0][0]
__________________________________________________________________________________________________
lstm_2 (LSTM) (None, 34, 256) 525312 embedding_1[0][0]
__________________________________________________________________________________________________
repeat_vector_1 (RepeatVector) (None, 34, 256) 0 dense_4[0][0]
__________________________________________________________________________________________________
time_distributed_1 (TimeDistrib (None, 34, 256) 65792 lstm_2[0][0]
__________________________________________________________________________________________________
concatenate_1 (Concatenate) (None, 34, 512) 0 repeat_vector_1[0][0]
time_distributed_1[0][0]
__________________________________________________________________________________________________
bidirectional_1 (Bidirectional) (None, 512) 1574912 concatenate_1[0][0]
__________________________________________________________________________________________________
dense_6 (Dense) (None, 256) 131328 bidirectional_1[0][0]
__________________________________________________________________________________________________
dense_7 (Dense) (None, 7579) 1947803 dense_6[0][0]
==================================================================================================
Total params: 6,709,915
Trainable params: 6,709,915
Non-trainable params: 0
__________________________________________________________________________________________________
None
# plot model
filename = 'models/rnn_model_2_inceptionv3.png'
plot(model, filename)
8. Fit Model
# Create sequences of images, input sequences and output words for an image
def create_sequences(tokenizer, max_length, captions_list, image):
# X1 : input for image features
# X2 : input for text features
# y : output word
X1, X2, y = list(), list(), list()
vocab_size = len(tokenizer.word_index) + 1
# Walk through each caption for the image
for caption in captions_list:
# Encode the sequence
seq = tokenizer.texts_to_sequences([caption])[0]
# Split one sequence into multiple X,y pairs
for i in range(1, len(seq)):
# Split into input and output pair
in_seq, out_seq = seq[:i], seq[i]
# Pad input sequence
in_seq = pad_sequences([in_seq], maxlen=max_length)[0]
# Encode output sequence
out_seq = to_categorical([out_seq], num_classes=vocab_size)[0]
# Store
X1.append(image)
X2.append(in_seq)
y.append(out_seq)
return X1, X2, y
# Data generator, intended to be used in a call to model.fit()
def data_generator(images, captions, tokenizer, max_length, batch_size, random_seed):
# Setting random seed for reproducibility of results
random.seed(random_seed)
# Image ids
image_ids = list(captions.keys())
_count=0
while True:
if _count >= len(image_ids):
# Generator exceeded or reached the end so restart it
_count = 0
# Batch list to store data
input_img_batch, input_sequence_batch, output_word_batch = list(), list(), list()
for i in range(_count, min(len(image_ids), _count+batch_size)):
# Retrieve the image id
image_id = image_ids[i]
# Retrieve the image features
image = images[image_id][0]
# Retrieve the captions list
captions_list = captions[image_id]
# Shuffle captions list
random.shuffle(captions_list)
input_img, input_sequence, output_word = create_sequences(tokenizer, max_length, captions_list, image)
# Add to batch
for j in range(len(input_img)):
input_img_batch.append(input_img[j])
input_sequence_batch.append(input_sequence[j])
output_word_batch.append(output_word[j])
_count = _count + batch_size
yield ([np.array(input_img_batch), np.array(input_sequence_batch)], np.array(output_word_batch))
# define parameters
num_of_epochs = 20
batch_size = 32
train_length = len(train_descriptions)
val_length = len(val_descriptions)
steps_train = train_length // batch_size
if train_length % batch_size != 0:
steps_train = steps_train+1
steps_val = val_length // batch_size
if val_length % batch_size != 0:
steps_val = steps_val+1
import random
# Setting random seed for reproducibility of results
random.seed('1000')
# Shuffle train data
ids_train = list(train_descriptions.keys())
random.shuffle(ids_train)
train_descriptions = {_id: train_descriptions[_id] for _id in ids_train}
# Create the train data generator
generator_train = data_generator(train_features, train_descriptions, tokenizer, max_length, batch_size, random_seed='1000')
# Create the validation data generator
generator_val = data_generator(val_features, val_descriptions, tokenizer, max_length, batch_size, random_seed='1000')
from keras.callbacks import ModelCheckpoint
from keras.callbacks import EarlyStopping
# define checkpoint callback
filepath = 'models/rnn_model_2_vgg16.h5'
checkpoint = ModelCheckpoint(filepath, monitor='val_loss', verbose=1, save_best_only=True, mode='min')
# define early stopping callback
early = EarlyStopping(patience=1, verbose=1)
# Fit
history = model.fit(generator_train,
epochs=num_of_epochs,
steps_per_epoch=steps_train,
validation_data=generator_val,
validation_steps=steps_val,
callbacks=[checkpoint, early],
verbose=1)
Epoch 1/20
188/188 [==============================] - 109s 527ms/step - loss: 6.2133 - val_loss: 5.3030
Epoch 00001: val_loss improved from inf to 5.30304, saving model to models/rnn_model_2_vgg16.h5
Epoch 2/20
188/188 [==============================] - 97s 517ms/step - loss: 5.2260 - val_loss: 4.7302
Epoch 00002: val_loss improved from 5.30304 to 4.73015, saving model to models/rnn_model_2_vgg16.h5
Epoch 3/20
188/188 [==============================] - 97s 514ms/step - loss: 4.6124 - val_loss: 4.2400
Epoch 00003: val_loss improved from 4.73015 to 4.23999, saving model to models/rnn_model_2_vgg16.h5
Epoch 4/20
188/188 [==============================] - 97s 514ms/step - loss: 4.1185 - val_loss: 3.9822
Epoch 00004: val_loss improved from 4.23999 to 3.98225, saving model to models/rnn_model_2_vgg16.h5
Epoch 5/20
188/188 [==============================] - 97s 516ms/step - loss: 3.7971 - val_loss: 3.8404
Epoch 00005: val_loss improved from 3.98225 to 3.84040, saving model to models/rnn_model_2_vgg16.h5
Epoch 6/20
188/188 [==============================] - 97s 517ms/step - loss: 3.5635 - val_loss: 3.7861
Epoch 00006: val_loss improved from 3.84040 to 3.78609, saving model to models/rnn_model_2_vgg16.h5
Epoch 7/20
188/188 [==============================] - 97s 516ms/step - loss: 3.3900 - val_loss: 3.7824
Epoch 00007: val_loss improved from 3.78609 to 3.78237, saving model to models/rnn_model_2_vgg16.h5
Epoch 8/20
188/188 [==============================] - 96s 512ms/step - loss: 3.2455 - val_loss: 3.7692
Epoch 00008: val_loss improved from 3.78237 to 3.76919, saving model to models/rnn_model_2_vgg16.h5
Epoch 9/20
188/188 [==============================] - 97s 516ms/step - loss: 3.1310 - val_loss: 3.7637
Epoch 00009: val_loss improved from 3.76919 to 3.76373, saving model to models/rnn_model_2_vgg16.h5
Epoch 10/20
188/188 [==============================] - 97s 514ms/step - loss: 3.0311 - val_loss: 3.7709
Epoch 00010: val_loss did not improve from 3.76373
Epoch 00010: early stopping
8.1. Model 1 VGG16
# using rnn_1 and vgg16
import matplotlib.pyplot as plt
for label in ["loss","val_loss"]:
plt.plot(history.history[label],label=label)
plt.legend()
plt.xlabel("epochs")
plt.ylabel("loss")
plt.savefig('models/rnn_1_vgg16_loss.png')
plt.show()
8.2. Model 1 InceptionV3
# using rnn_1 and inceptionv3
import matplotlib.pyplot as plt
for label in ["loss","val_loss"]:
plt.plot(history.history[label],label=label)
plt.legend()
plt.xlabel("epochs")
plt.ylabel("loss")
plt.savefig('models/rnn_1_inceptionv3_loss.png')
plt.show()
8.3. Model 2 VGG16
# using rnn_2 and vgg16
import matplotlib.pyplot as plt
for label in ["loss","val_loss"]:
plt.plot(history.history[label],label=label)
plt.legend()
plt.xlabel("epochs")
plt.ylabel("loss")
plt.savefig('models/rnn_2_vgg16_loss.png')
plt.show()
8.4. Model 2 InceptionV3
# using rnn_2 and inceptionv3
import matplotlib.pyplot as plt
for label in ["loss","val_loss"]:
plt.plot(history.history[label],label=label)
plt.legend()
plt.xlabel("epochs")
plt.ylabel("loss")
plt.savefig('models/rnn_2_inceptionv3_loss.png')
plt.show()
9. Evaluate Model
9.1. Sampling
# generate a description for an image
def generate_desc(model, tokenizer, photo, max_length):
# seed the generation process
in_text = '<start>'
# iterate over the whole length of the sequence
for i in range(max_length):
# integer encode input sequence
sequence = tokenizer.texts_to_sequences([in_text])[0]
# pad input
sequence = pad_sequences([sequence], maxlen=max_length)
# predict next word
yhat = model.predict([photo,sequence], verbose=0)
# convert probability to integer
yhat = argmax(yhat)
# map integer to word
word = tokenizer.index_word[yhat]
# stop if we cannot map the word
if word is None:
break
# append as input for generating the next word
in_text += ' ' + word
# stop if we predict the end of the sequence
if word == '<end>':
break
return in_text
9.2. Beam Search
# generate a description for an image using beam search
def generate_desc_beam_search(model, tokenizer, photo, max_length, beam_index=3):
# seed the generation process
in_text = [['<start>', 0.0]]
# iterate over the whole length of the sequence
for i in range(max_length):
temp = []
for s in in_text:
# integer encode input sequence
sequence = tokenizer.texts_to_sequences([s[0]])[0]
# pad input
sequence = pad_sequences([sequence], maxlen=max_length)
# predict next words
preds = model.predict([photo,sequence], verbose=0)
word_preds = argsort(preds[0])[-beam_index:]
# get top predictions
for w in word_preds:
next_cap, prob = s[0][:], s[1]
# map integer to word
word = tokenizer.index_word[w]
next_cap += ' ' + word
prob += preds[0][w]
temp.append([next_cap, prob])
in_text = temp
# sorting according to the probabilities
in_text = sorted(in_text, reverse=False, key=lambda l: l[1])
# getting the top words
in_text = in_text[-beam_index:]
# get last (best) caption text
in_text = in_text[-1][0]
caption_list = []
# remove leftover <end>
for w in in_text.split():
caption_list.append(w)
if w == '<end>':
break
# convert list to string
caption = ' '.join(caption_list)
return caption
9.3. BLEU
def calculate_scores(actual, predicted):
# calculate BLEU score
smooth = SmoothingFunction().method4
bleu1 = corpus_bleu(actual, predicted, weights=(1.0, 0, 0, 0), smoothing_function=smooth)*100
bleu2 = corpus_bleu(actual, predicted, weights=(0.5, 0.5, 0, 0), smoothing_function=smooth)*100
bleu3 = corpus_bleu(actual, predicted, weights=(0.3, 0.3, 0.3, 0), smoothing_function=smooth)*100
bleu4 = corpus_bleu(actual, predicted, weights=(0.25, 0.25, 0.25, 0.25), smoothing_function=smooth)*100
print('BLEU-1: %f' % bleu1)
print('BLEU-2: %f' % bleu2)
print('BLEU-3: %f' % bleu3)
print('BLEU-4: %f' % bleu4)
# evaluate the skill of the model
def evaluate_model(model, descriptions, features, tokenizer, max_length):
actual, predicted = list(), list()
# step over the whole set
for key, desc_list in tqdm(descriptions.items(), position=0, leave=True):
# generate description
yhat = generate_desc(model, tokenizer, features[key], max_length)
# store actual and predicted
references = [d.split() for d in desc_list]
actual.append(references)
predicted.append(yhat.split())
print('Sampling:')
calculate_scores(actual, predicted)
# evaluate the skill of the model
def evaluate_model_beam_search(model, descriptions, features, tokenizer, max_length, beam_index=3):
actual, predicted = list(), list()
# step over the whole set
for key, desc_list in tqdm(descriptions.items(), position=0, leave=True):
# generate description beam search
yhat = generate_desc_beam_search(model, tokenizer, features[key], max_length, beam_index)
# store actual and predicted
references = [d.split() for d in desc_list]
actual.append(references)
predicted.append(yhat.split())
print('Beam Search k=%d:' % beam_index)
calculate_scores(actual, predicted)
9.3.1. Model 1 VGG16
# load the model
filename = 'models/rnn_model_1_vgg16.h5'
model = load_model(filename)
# evaluate model
evaluate_model(model, test_descriptions, test_features, tokenizer, max_length)
100%|██████████| 1000/1000 [07:15<00:00, 2.30it/s]
Sampling:
BLEU-1: 54.114390
BLEU-2: 33.489586
BLEU-3: 24.785442
BLEU-4: 13.031321
9.3.2. Model 1 InceptionV3
# load the model
filename = 'models/rnn_model_1_inceptionv3.h5'
model = load_model(filename)
# evaluate model
evaluate_model(model, test_descriptions, test_features, tokenizer, max_length)
100%|██████████| 1000/1000 [06:48<00:00, 2.45it/s]
Sampling:
BLEU-1: 57.458612
BLEU-2: 34.077377
BLEU-3: 24.426512
BLEU-4: 12.300981
9.3.3. Model 2 VGG16
# load the model
filename = 'models/rnn_model_2_vgg16.h5'
model = load_model(filename)
# evaluate model
evaluate_model(model, test_descriptions, test_features, tokenizer, max_length)
Sampling:
BLEU-1: 0.595796
BLEU-2: 0.369997
BLEU-3: 0.272431
BLEU-4: 0.144684
9.3.4. Model 2 InceptionV3
# load the model
filename = 'models/rnn_model_2_inceptionv3.h5'
model = load_model(filename)
# evaluate model
evaluate_model(model, test_descriptions, test_features, tokenizer, max_length)
100%|██████████| 1000/1000 [09:01<00:00, 1.85it/s]
Sampling:
BLEU-1: 0.587866
BLEU-2: 0.366849
BLEU-3: 0.270134
BLEU-4: 0.144244
# evaluate model beam_search
evaluate_model_beam_search(model, test_descriptions, test_features, tokenizer, max_length, beam_index=3)
55%|█████▌ | 552/1000 [50:39<40:36, 5.44s/it]
10. Generate Captions
def clean_caption(caption):
# split caption words
caption_list = caption.split()
# remove <start> and <end>
caption_list = caption_list[1:len(caption_list)-1]
# convert list to string
caption = ' '.join(caption_list)
return caption
def generate_captions(model, descriptions, features, image_size, count):
c = 0
for key, desc_list in descriptions.items():
# load an image from file
filename = 'Flickr8k_Dataset/' + key + '.jpg'
#diplay image
display(Image(filename))
# print original descriptions
for i, desc in enumerate(desc_list):
print('Original ' + str(i+1) + ': ' + clean_caption(desc_list[i]))
# generate descriptions
desc = generate_desc(model, tokenizer, features[key], max_length)
desc_beam_3 = generate_desc_beam_search(model, tokenizer, features[key], max_length, beam_index=3)
desc_beam_5 = generate_desc_beam_search(model, tokenizer, features[key], max_length, beam_index=5)
# calculate BLEU-1 scores
references = [d.split() for d in desc_list]
smooth = SmoothingFunction().method4
desc_bleu = sentence_bleu(references, desc.split(), weights=(1.0, 0, 0, 0), smoothing_function=smooth)*100
desc_beam_3_bleu = sentence_bleu(references, desc_beam_3.split(), weights=(1.0, 0, 0, 0), smoothing_function=smooth)*100
desc_beam_5_bleu = sentence_bleu(references, desc_beam_5.split(), weights=(1.0, 0, 0, 0), smoothing_function=smooth)*100
# print descriptions with scores
print('Sampling (BLEU-1: %f): %s' % (desc_bleu, clean_caption(desc)))
print('Beam Search k=3 (BLEU-1: %f): %s' % (desc_beam_3_bleu, clean_caption(desc_beam_3)))
print('Beam Search k=5 (BLEU-1: %f): %s' % (desc_beam_5_bleu, clean_caption(desc_beam_5)))
c += 1
if c == count:
break
# load the tokenizer
tokenizer = load(open('files/tokenizer.pkl', 'rb'))
10.1. Model 1 VGG16
# load the model
filename = 'models/rnn_model_1_vgg16.h5'
rnn_model = load_model(filename)
# using rnn_2 and vgg16 model
generate_captions(rnn_model, test_descriptions, test_features, 224, 10)
Original 1: blond woman in blue shirt appears to wait for ride
Original 2: blond woman is on the street hailing taxi
Original 3: woman is signaling is to traffic as seen from behind
Original 4: woman with blonde hair wearing blue tube top is waving on the side of the street
Original 5: the woman in the blue dress is holding out her arm at oncoming traffic
Sampling (BLEU-1: 69.230769): man in blue shirt and sunglasses is sitting on the street
Beam Search k=3 (BLEU-1: 60.000000): man and woman are sitting on the street
Beam Search k=5 (BLEU-1: 50.000000): man and woman are sitting on the middle of camera
Original 1: boy in his blue swim shorts at the beach
Original 2: boy smiles for the camera at beach
Original 3: young boy in swimming trunks is walking with his arms outstretched on the beach
Original 4: children playing on the beach
Original 5: the boy is playing on the shore of an ocean
Sampling (BLEU-1: 81.818182): boy in blue shirt is standing on the beach
Beam Search k=3 (BLEU-1: 83.333333): little boy in blue shirt is walking on the beach
Beam Search k=5 (BLEU-1: 83.333333): little boy in blue shirt is walking on the beach
Original 1: lady and man with no shirt sit on dock
Original 2: man and woman are sitting on dock together
Original 3: man and woman sitting on dock
Original 4: man and woman sitting on deck next to lake
Original 5: shirtless man and woman sitting on dock
Sampling (BLEU-1: 60.000000): man in blue shirt is sitting on bench
Beam Search k=3 (BLEU-1: 36.363636): young boy is sitting on bench in the park
Beam Search k=5 (BLEU-1: 41.666667): the man is sitting on bench in front of park
Original 1: closeup of white dog that is laying its head on its paws
Original 2: large white dog lying on the floor
Original 3: white dog has its head on the ground
Original 4: white dog is resting its head on tiled floor with its eyes open
Original 5: white dog rests its head on the patio bricks
Sampling (BLEU-1: 66.666667): white dog is running through the air
Beam Search k=3 (BLEU-1: 42.736665): white and white and white dog is standing in its mouth
Beam Search k=5 (BLEU-1: 42.736665): white and white and white dog is standing in its mouth
Original 1: boy with toy gun
Original 2: little boy in orange shorts playing with toy
Original 3: young boy with his foot outstretched aims toy at the camera in front of fireplace
Original 4: young child plays with his new lightup toy
Original 5: boy with toy gun pointed at the camera
Sampling (BLEU-1: 45.454545): boy in blue shirt is sitting on the ground
Beam Search k=3 (BLEU-1: 50.000000): young boy in blue shirt is sitting on the water
Beam Search k=5 (BLEU-1: 50.000000): little boy in blue shirt is sitting on the beach
Original 1: black dog jumping to catch rope toy
Original 2: black dog playing fetch with ball of rope
Original 3: black dog pounces to get rope toy
Original 4: black dog running after his rope toy
Original 5: large black dog is playing in grassy yard
Sampling (BLEU-1: 66.666667): black dog is running through the grass
Beam Search k=3 (BLEU-1: 66.666667): black dog is running through the grass
Beam Search k=5 (BLEU-1: 44.124845): black dog runs through the grass
Original 1: little tan dog with large ears running through the grass
Original 2: playful dog is running through the grass
Original 3: small dogs ears stick up as it runs in the grass
Original 4: the small dog is running across the lawn
Original 5: this is small beige dog running through grassy field
Sampling (BLEU-1: 77.777778): two dogs are running in the grass
Beam Search k=3 (BLEU-1: 66.187268): two dogs play in the grass
Beam Search k=5 (BLEU-1: 70.000000): brown and white dog running through the grass
Original 1: baby in white garment holds flag with crescent moon and star
Original 2: baby is holding small black flag with moon and star on it
Original 3: baby wearing white gown waves muslim flag
Original 4: little toddler dressed in white is smiling while lady helps him wave flag
Original 5: baby in white outfit holding black and white flag
Sampling (BLEU-1: 33.333333): two children are sitting on the camera
Beam Search k=3 (BLEU-1: 41.666667): young boy in blue shirt is sitting on the camera
Beam Search k=5 (BLEU-1: 41.666667): young boy in blue shirt is sitting on the camera
Original 1: brown and white dog stands outside while it snows
Original 2: dog is looking at something near the water
Original 3: furry dog attempts to dry itself by shaking the water off its coat
Original 4: white and brown dog shaking its self dry
Original 5: the large brown and white dog shakes off water
Sampling (BLEU-1: 69.598614): white dog is running through the water
Beam Search k=3 (BLEU-1: 11.428571): white and white and white and white and white and white and white and white and white and white and white and white and white and white and white and white and white
Beam Search k=5 (BLEU-1: 11.428571): black and white and white and white and white and white and white and white and white and white and white and white and white and white and white and white and white
Original 1: boy is jumping on bed
Original 2: boy jumped up from the green bed
Original 3: boy jumps from one bed to another
Original 4: small child is jumping on bed
Original 5: the boy in blue shorts is bouncing on the bed
Sampling (BLEU-1: 75.000000): young boy in blue shirt is sitting on the bed
Beam Search k=3 (BLEU-1: 77.777778): the little boy is sitting on bed
Beam Search k=5 (BLEU-1: 75.000000): little boy is sitting on bed
10.2. Model 1 InceptionV3
# load the model
filename = 'models/rnn_model_1_inceptionv3.h5'
rnn_model = load_model(filename)
# using rnn_2 and inceptionv3 model
generate_captions(rnn_model, test_descriptions, test_features, 299, 10)
Original 1: blond woman in blue shirt appears to wait for ride
Original 2: blond woman is on the street hailing taxi
Original 3: woman is signaling is to traffic as seen from behind
Original 4: woman with blonde hair wearing blue tube top is waving on the side of the street
Original 5: the woman in the blue dress is holding out her arm at oncoming traffic
Sampling (BLEU-1: 56.250000): man in blue jacket and blue hat is holding his head on the street
Beam Search k=3 (BLEU-1: 62.500000): man in blue shirt and blue hat is holding his arms on the street
Beam Search k=5 (BLEU-1: 58.333333): young boy in blue jacket is sitting on the beach
Original 1: boy in his blue swim shorts at the beach
Original 2: boy smiles for the camera at beach
Original 3: young boy in swimming trunks is walking with his arms outstretched on the beach
Original 4: children playing on the beach
Original 5: the boy is playing on the shore of an ocean
Sampling (BLEU-1: 56.130419): young boy in blue bathing suit is holding his head in the water
Beam Search k=3 (BLEU-1: 61.538462): young boy in blue bathing suit is running in the water
Beam Search k=5 (BLEU-1: 69.230769): young boy in blue bathing suit is playing in the water
Original 1: lady and man with no shirt sit on dock
Original 2: man and woman are sitting on dock together
Original 3: man and woman sitting on dock
Original 4: man and woman sitting on deck next to lake
Original 5: shirtless man and woman sitting on dock
Sampling (BLEU-1: 58.333333): man in blue shirt and jeans is sitting on bench
Beam Search k=3 (BLEU-1: 38.461538): the man is sitting on bench in front of the ocean
Beam Search k=5 (BLEU-1: 38.461538): the man is sitting on bench in front of white building
Original 1: closeup of white dog that is laying its head on its paws
Original 2: large white dog lying on the floor
Original 3: white dog has its head on the ground
Original 4: white dog is resting its head on tiled floor with its eyes open
Original 5: white dog rests its head on the patio bricks
Sampling (BLEU-1: 55.156056): dog is running through the air
Beam Search k=3 (BLEU-1: 54.545455): white and white dog is running through the woods
Beam Search k=5 (BLEU-1: 50.000000): white and white dog runs through the woods
Original 1: boy with toy gun
Original 2: little boy in orange shorts playing with toy
Original 3: young boy with his foot outstretched aims toy at the camera in front of fireplace
Original 4: young child plays with his new lightup toy
Original 5: boy with toy gun pointed at the camera
Sampling (BLEU-1: 29.411765): man in blue shirt is sitting on the edge of the edge of the water
Beam Search k=3 (BLEU-1: 23.337955): the man is sitting on the edge of the edge of the water
Beam Search k=5 (BLEU-1: 54.545455): young boy is sitting on the edge of water
Original 1: black dog jumping to catch rope toy
Original 2: black dog playing fetch with ball of rope
Original 3: black dog pounces to get rope toy
Original 4: black dog running after his rope toy
Original 5: large black dog is playing in grassy yard
Sampling (BLEU-1: 66.666667): black dog is running through the water
Beam Search k=3 (BLEU-1: 55.156056): black dog running through the water
Beam Search k=5 (BLEU-1: 44.124845): black dog runs through the water
Original 1: little tan dog with large ears running through the grass
Original 2: playful dog is running through the grass
Original 3: small dogs ears stick up as it runs in the grass
Original 4: the small dog is running across the lawn
Original 5: this is small beige dog running through grassy field
Sampling (BLEU-1: 88.888889): white dog is running through the grass
Beam Search k=3 (BLEU-1: 72.727273): brown and white dog is running through the grass
Beam Search k=5 (BLEU-1: 70.000000): brown and white dog runs through the grass
Original 1: baby in white garment holds flag with crescent moon and star
Original 2: baby is holding small black flag with moon and star on it
Original 3: baby wearing white gown waves muslim flag
Original 4: little toddler dressed in white is smiling while lady helps him wave flag
Original 5: baby in white outfit holding black and white flag
Sampling (BLEU-1: 45.454545): man in blue shirt is sitting on the camera
Beam Search k=3 (BLEU-1: 45.454545): young boy in blue shirt is sitting on bed
Beam Search k=5 (BLEU-1: 45.454545): young boy in blue shirt is sitting on bed
Original 1: brown and white dog stands outside while it snows
Original 2: dog is looking at something near the water
Original 3: furry dog attempts to dry itself by shaking the water off its coat
Original 4: white and brown dog shaking its self dry
Original 5: the large brown and white dog shakes off water
Sampling (BLEU-1: 69.598614): white dog is swimming in the water
Beam Search k=3 (BLEU-1: 80.000000): brown and white dog swims in the water
Beam Search k=5 (BLEU-1: 80.000000): brown and white dog swims in the water
Original 1: boy is jumping on bed
Original 2: boy jumped up from the green bed
Original 3: boy jumps from one bed to another
Original 4: small child is jumping on bed
Original 5: the boy in blue shorts is bouncing on the bed
Sampling (BLEU-1: 70.000000): man in blue shirt is sitting on bed
Beam Search k=3 (BLEU-1: 70.000000): man in blue shirt is sitting on bed
Beam Search k=5 (BLEU-1: 62.500000): little girl is sitting on bed
10.3. Model 2 VGG16
# load the model
filename = 'models/rnn_model_2_vgg16.h5'
rnn_model = load_model(filename)
# using rnn_2 and vgg16 model
generate_captions(rnn_model, test_descriptions, test_features, 224, 10)
Original 1: blond woman in blue shirt appears to wait for ride
Original 2: blond woman is on the street hailing taxi
Original 3: woman is signaling is to traffic as seen from behind
Original 4: woman with blonde hair wearing blue tube top is waving on the side of the street
Original 5: the woman in the blue dress is holding out her arm at oncoming traffic
Sampling (BLEU-1: 58.410059): man is sitting on the street
Beam Search k=3 (BLEU-1: 69.598614): the man is sitting on the street
Beam Search k=5 (BLEU-1: 69.598614): the man is sitting on the street
Original 1: boy in his blue swim shorts at the beach
Original 2: boy smiles for the camera at beach
Original 3: young boy in swimming trunks is walking with his arms outstretched on the beach
Original 4: children playing on the beach
Original 5: the boy is playing on the shore of an ocean
Sampling (BLEU-1: 85.714286): young boy in the sand
Beam Search k=3 (BLEU-1: 85.714286): young boy in the sand
Beam Search k=5 (BLEU-1: 85.714286): young boy in the sand
Original 1: lady and man with no shirt sit on dock
Original 2: man and woman are sitting on dock together
Original 3: man and woman sitting on dock
Original 4: man and woman sitting on deck next to lake
Original 5: shirtless man and woman sitting on dock
Sampling (BLEU-1: 37.500000): man is jumping off of skateboard
Beam Search k=3 (BLEU-1: 37.151910): man jumps off of skateboard
Beam Search k=5 (BLEU-1: 55.555556): man on skateboard is sitting on rail
Original 1: closeup of white dog that is laying its head on its paws
Original 2: large white dog lying on the floor
Original 3: white dog has its head on the ground
Original 4: white dog is resting its head on tiled floor with its eyes open
Original 5: white dog rests its head on the patio bricks
Sampling (BLEU-1: 33.333333): two dogs are playing in the grass
Beam Search k=3 (BLEU-1: 55.156056): white dog looks at the steps
Beam Search k=5 (BLEU-1: 66.187268): white dog looks at the ground
Original 1: boy with toy gun
Original 2: little boy in orange shorts playing with toy
Original 3: young boy with his foot outstretched aims toy at the camera in front of fireplace
Original 4: young child plays with his new lightup toy
Original 5: boy with toy gun pointed at the camera
Sampling (BLEU-1: 50.000000): the little boy is sitting on the floor
Beam Search k=3 (BLEU-1: 60.000000): the little boy is sitting in the grass
Beam Search k=5 (BLEU-1: 60.000000): the little boy is sitting in the grass
Original 1: black dog jumping to catch rope toy
Original 2: black dog playing fetch with ball of rope
Original 3: black dog pounces to get rope toy
Original 4: black dog running after his rope toy
Original 5: large black dog is playing in grassy yard
Sampling (BLEU-1: 66.666667): black dog is running on the grass
Beam Search k=3 (BLEU-1: 70.000000): the black dog is running in the grass
Beam Search k=5 (BLEU-1: 54.545455): black dog and black dog running in the grass
Original 1: little tan dog with large ears running through the grass
Original 2: playful dog is running through the grass
Original 3: small dogs ears stick up as it runs in the grass
Original 4: the small dog is running across the lawn
Original 5: this is small beige dog running through grassy field
Sampling (BLEU-1: 44.124845): two dogs are playing with ball
Beam Search k=3 (BLEU-1: 60.000000): two brown dogs are playing in the grass
Beam Search k=5 (BLEU-1: 66.666667): brown dog is playing with tennis ball in the grass
Original 1: baby in white garment holds flag with crescent moon and star
Original 2: baby is holding small black flag with moon and star on it
Original 3: baby wearing white gown waves muslim flag
Original 4: little toddler dressed in white is smiling while lady helps him wave flag
Original 5: baby in white outfit holding black and white flag
Sampling (BLEU-1: 30.000000): two young girls are playing with each other
Beam Search k=3 (BLEU-1: 30.000000): two young girls are sitting in the room
Beam Search k=5 (BLEU-1: 22.222222): two young girls are posing for picture
Original 1: brown and white dog stands outside while it snows
Original 2: dog is looking at something near the water
Original 3: furry dog attempts to dry itself by shaking the water off its coat
Original 4: white and brown dog shaking its self dry
Original 5: the large brown and white dog shakes off water
Sampling (BLEU-1: 34.227808): dog with its mouth
Beam Search k=3 (BLEU-1: 59.655954): black and white dog in the sand
Beam Search k=5 (BLEU-1: 59.655954): black and white dog in the snow
Original 1: boy is jumping on bed
Original 2: boy jumped up from the green bed
Original 3: boy jumps from one bed to another
Original 4: small child is jumping on bed
Original 5: the boy in blue shorts is bouncing on the bed
Sampling (BLEU-1: 90.000000): the little boy is jumping on the bed
Beam Search k=3 (BLEU-1: 77.777778): the little boy is in the tunnel
Beam Search k=5 (BLEU-1: 77.777778): the little boy is in the air
10.4. Model 2 InceptionV3
# load the model
filename = 'models/rnn_model_2_inceptionv3.h5'
rnn_model = load_model(filename)
# using rnn_2 and inceptionv3 model
generate_captions(rnn_model, test_descriptions, test_features, 299, 10)
Original 1: blond woman in blue shirt appears to wait for ride
Original 2: blond woman is on the street hailing taxi
Original 3: woman is signaling is to traffic as seen from behind
Original 4: woman with blonde hair wearing blue tube top is waving on the side of the street
Original 5: the woman in the blue dress is holding out her arm at oncoming traffic
Sampling (BLEU-1: 0.500000): the boy is wearing red hat and is wearing helmet
Beam Search k=3 (BLEU-1: 0.600000): the boy is wearing red shirt and helmet
Beam Search k=5 (BLEU-1: 0.500000): young boy wearing helmet and blue shirt is wearing helmet
Original 1: boy in his blue swim shorts at the beach
Original 2: boy smiles for the camera at beach
Original 3: young boy in swimming trunks is walking with his arms outstretched on the beach
Original 4: children playing on the beach
Original 5: the boy is playing on the shore of an ocean
Sampling (BLEU-1: 0.636364): boy in blue shirt is standing in the water
Beam Search k=3 (BLEU-1: 0.714286): little boy in swimming pool
Beam Search k=5 (BLEU-1: 0.857143): young boy in swimming pool
Original 1: lady and man with no shirt sit on dock
Original 2: man and woman are sitting on dock together
Original 3: man and woman sitting on dock
Original 4: man and woman sitting on deck next to lake
Original 5: shirtless man and woman sitting on dock
Sampling (BLEU-1: 0.600000): man in blue shirt is sitting on bench
Beam Search k=3 (BLEU-1: 0.400000): the man in the blue shirt is sitting on the top of rock
Beam Search k=5 (BLEU-1: 0.400000): the man in the blue shirt is sitting on the top of rock
Original 1: closeup of white dog that is laying its head on its paws
Original 2: large white dog lying on the floor
Original 3: white dog has its head on the ground
Original 4: white dog is resting its head on tiled floor with its eyes open
Original 5: white dog rests its head on the patio bricks
Sampling (BLEU-1: 0.500000): dog is jumping over log in the air
Beam Search k=3 (BLEU-1: 0.600000): the white dog is running through the snow
Beam Search k=5 (BLEU-1: 0.583333): the white dog is in the middle of the snow
Original 1: boy with toy gun
Original 2: little boy in orange shorts playing with toy
Original 3: young boy with his foot outstretched aims toy at the camera in front of fireplace
Original 4: young child plays with his new lightup toy
Original 5: boy with toy gun pointed at the camera
Sampling (BLEU-1: 0.298280): boy is holding baby on her face
Beam Search k=3 (BLEU-1: 0.666667): the little boy is standing in front of some water
Beam Search k=5 (BLEU-1: 0.461210): the little boy in the blue shirt is playing with brown dog
Original 1: black dog jumping to catch rope toy
Original 2: black dog playing fetch with ball of rope
Original 3: black dog pounces to get rope toy
Original 4: black dog running after his rope toy
Original 5: large black dog is playing in grassy yard
Sampling (BLEU-1: 0.666667): black dog is jumping into the air
Beam Search k=3 (BLEU-1: 0.600000): the black dog is running through the grass
Beam Search k=5 (BLEU-1: 0.600000): the black dog is running through the grass
Original 1: little tan dog with large ears running through the grass
Original 2: playful dog is running through the grass
Original 3: small dogs ears stick up as it runs in the grass
Original 4: the small dog is running across the lawn
Original 5: this is small beige dog running through grassy field
Sampling (BLEU-1: 0.882497): dog is running through the grass
Beam Search k=3 (BLEU-1: 0.666667): white dog is playing with green ball in the grass
Beam Search k=5 (BLEU-1: 0.666667): white dog is playing with red ball in the grass
Original 1: baby in white garment holds flag with crescent moon and star
Original 2: baby is holding small black flag with moon and star on it
Original 3: baby wearing white gown waves muslim flag
Original 4: little toddler dressed in white is smiling while lady helps him wave flag
Original 5: baby in white outfit holding black and white flag
Sampling (BLEU-1: 0.454545): boy with his head is holding up his head
Beam Search k=3 (BLEU-1: 0.416667): young boy is holding spoon in front of brick wall
Beam Search k=5 (BLEU-1: 0.545455): little boy in blue shirt is playing with bubble
Original 1: brown and white dog stands outside while it snows
Original 2: dog is looking at something near the water
Original 3: furry dog attempts to dry itself by shaking the water off its coat
Original 4: white and brown dog shaking its self dry
Original 5: the large brown and white dog shakes off water
Sampling (BLEU-1: 0.584101): dog is swimming in the water
Beam Search k=3 (BLEU-1: 0.727273): white and white dog is swimming in the water
Beam Search k=5 (BLEU-1: 0.700000): brown and white dog swims in shallow water
Original 1: boy is jumping on bed
Original 2: boy jumped up from the green bed
Original 3: boy jumps from one bed to another
Original 4: small child is jumping on bed
Original 5: the boy in blue shorts is bouncing on the bed
Sampling (BLEU-1: 0.500000): two children are sitting on bed
Beam Search k=3 (BLEU-1: 0.664073): young boy in blue shirt is sitting on bed
Beam Search k=5 (BLEU-1: 0.664073): little boy in blue shirt is sitting on bed
Julen Etxaniz
Estudiante de Doctorado en Análisis y Procesamiento del Lenguaje
Estudiante de Doctorado en Análisis y Procesamiento del Lenguaje en HiTZ Center IXA Group (UPV/EHU). Trabajando en mejorar los modelos de lenguaje para idiomas con pocos recursos. Graduado en Ingeniería Informática con especialidad en Ingeniería del Software. Máster en Análisis y Procesamiento del Lenguaje.
