#!/usr/bin/env python
# -*- coding: utf-8 -*-
'''
Sample code to generate caption
'''
import numpy as np
from image_reader import Image_reader
from caption_generator import Caption_generator
#Instantiate image_reader with GoogleNet mean image
mean_image = np.array([104, 117, 123]).reshape((3, 1, 1))
image_reader = Image_reader(mean=mean_image)
#Instantiate caption generator
caption_model_place = '../models/caption_model.chainer'
cnn_model_place = '../data/bvlc_googlenet_caffe_chainer.pkl'
index2word_place = '../work/index2token.pkl'
caption_generator = Caption_generator(caption_model_place=caption_model_place,
cnn_model_place=cnn_model_place,
index2word_place=index2word_place)
#The preparation is done
#Let's ganarate caption for a image
#First, read an image as numpy array
image_file_path = '../images/test_image.jpg'
image = image_reader.read(image_file_path)
#Next, put the image into caption generator
#The output structure is
# [caption,caption,caption,...]
parser.add_argument("-v", "--vocab",default=index2word_file, type=str, help=u" vocaburary file")
parser.add_argument("-i", "--image",default=image_file_name, type=str, help=u"a image that you want to generate capiton ")
parser.add_argument("-b", "--beam",default=beamsize, type=int, help=u"a image that you want to generate capiton ")
args = parser.parse_args()
gpu_id=args.gpu
model_place= args.model
index2word_file = args.vocab
image_file_name = args.image
caffe_model_place = args.caffe
beamsize = args.beam
#Instantiate image_reader with GoogleNet mean image
mean_image = np.array([104, 117, 123]).reshape((3,1,1))#GoogleNet Mean
image_reader=Image_reader(mean=mean_image)
#Instantiate caption generator
caption_generator=Caption_generator(caption_model_place=model_place,cnn_model_place=caffe_model_place,index2word_place=index2word_file,beamsize=beamsize,gpu_id=gpu_id)
#Read Image
image=image_reader.read(image_file_name)
#Generate Catpion
captions=caption_generator.generate(image)
#print it
for caption in captions:
sentence=caption['sentence']
probability=caption['probability']
print " ".join(sentence),probability
#!/usr/bin/env python # -*- coding: utf-8 -*- ''' Sample code to generate caption ''' import numpy as np from image_reader import Image_reader from caption_generator import Caption_generator #Instantiate image_reader with GoogleNet mean image mean_image = np.array([104, 117, 123]).reshape((3,1,1)) image_reader=Image_reader(mean=mean_image) #Instantiate caption generator caption_model_place='../models/caption_model.chainer' cnn_model_place='../data/bvlc_googlenet_caffe_chainer.pkl' index2word_place='../work/index2token.pkl' caption_generator=Caption_generator(caption_model_place=caption_model_place,cnn_model_place=cnn_model_place,index2word_place=index2word_place) #The preparation is done #Let's ganarate caption for a image #First, read an image as numpy array image_file_path='../images/test_image.jpg' image=image_reader.read(image_file_path) #Next, put the image into caption generator #The output structure is
"--beam",
default=beamsize,
type=int,
help=u"a image that you want to generate capiton ")
args = parser.parse_args()
gpu_id = args.gpu
model_place = args.model
index2word_file = args.vocab
image_file_name = args.image
caffe_model_place = args.caffe
beamsize = args.beam
#Instantiate image_reader with GoogleNet mean image
mean_image = np.array([104, 117, 123]).reshape((3, 1, 1)) #GoogleNet Mean
image_reader = Image_reader(mean=mean_image)
#Instantiate caption generator
caption_generator = Caption_generator(caption_model_place=model_place,
cnn_model_place=caffe_model_place,
index2word_place=index2word_file,
beamsize=beamsize,
gpu_id=gpu_id)
#Read Image
image = image_reader.read(image_file_name)
#Generate Catpion
captions = caption_generator.generate(image)
#print it
