def extract_features(path, model_type):
if model_type == 'inceptionv3':
from keras.applications.inception_v3 import preprocess_input
target_size = (299, 299)
elif model_type == 'vgg16':
from keras.applications.vgg16 import preprocess_input
target_size = (224, 224)
# Get CNN Model from model.py
model = CNNModel(model_type)
features = dict()
# Extract features from each photo
for name in tqdm(os.listdir(path)):
# Loading and resizing image
filename = path + name
image = load_img(filename, target_size=target_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 CNN Model model
image = preprocess_input(image)
# Pass image into model to get encoded features
feature = model.predict(image, verbose=0)
# Store encoded features for the image
image_id = name.split('.')[0]
features[image_id] = feature
return features
def __init__(self):
tokenizer_path = get_absolute_path(config['tokenizer_path'])
with open(tokenizer_path, 'rb') as f:
self._tokenizer = load(f)
self._max_length = config['max_length']
self._caption_model = load_model(
get_absolute_path(config['model_load_path']))
self._image_model = CNNModel(config['model_type'])
def run_captioning(image_file):
# import datetime
# today = datetime.date.today()
# image_file = '/home/lab02/imgdesc/django/media/' + '{:/%Y%m/%d/}'.format(today) + image_file_name
# Load the tokenizer
tokenizer_path = config['tokenizer_path']
tokenizer = load(open(tokenizer_path, 'rb'))
# Max sequence length (from training)
max_length = config['max_length']
# Load the model
caption_model = load_model(config['model_load_path'])
image_model = CNNModel(config['model_type'])
# Load and prepare the image
# if (image_file.split('.')[1] == 'jpg' or image_file.split('.')[1] == 'jpeg'):
try:
# Encode image using CNN Model
image = extract_features(image_file, image_model, config['model_type'])
except:
return (config['errmsg_imgopen'] + '.')
try:
if len(image) == 0:
return (config['errmsg_imgopen'] + '')
# Generate caption using Decoder RNN Model + BEAM search
generated_caption = generate_caption_beam_search(
caption_model,
tokenizer,
image,
max_length,
beam_index=config['beam_search_k'])
except:
return (config['errmsg_imgopen'] + '..')
try:
# Remove startseq and endseq
desc_en = generated_caption.split()[1].capitalize()
for x in generated_caption.split()[2:len(generated_caption.split()) -
1]:
desc_en = desc_en + ' ' + x
desc_en += '.'
except:
return (config['errmsg_imgopen'] + '...')
try:
# show captions
print('BEAM Search with k=', config['beam_search_k'])
print(desc_en)
return (desc_en)
# print(desc_ko,'\n\n')
except:
return (config['errmsg_imgopen'] + '....')
image = preprocess_input(image)
# Pass image into model to get encoded features
features = model.predict(image, verbose=0)
return features
# Load the tokenizer
tokenizer_path = config['tokenizer_path']
tokenizer = load(open(tokenizer_path, 'rb'))
# Max sequence length (from training)
max_length = config['max_length']
# Load the model
caption_model = load_model(config['model_load_path'])
image_model = CNNModel(config['model_type'])
# Load and prepare the image
for image_file in os.listdir(config['test_data_path']):
if(image_file.split('--')[0]=='output'):
continue
if(image_file.split('.')[1]=='jpg' or image_file.split('.')[1]=='jpeg'):
print('Generating caption for {}'.format(image_file))
# Encode image using CNN Model
image = extract_features(config['test_data_path']+image_file, image_model, config['model_type'])
# Generate caption using Decoder RNN Model + BEAM search
generated_caption = generate_caption_beam_search(caption_model, tokenizer, image, max_length, beam_index=config['beam_search_k'])
# Remove startseq and endseq
caption = 'Caption: ' + generated_caption.split()[1].capitalize()
for x in generated_caption.split()[2:len(generated_caption.split())-1]: