loading_script.py

import os
import sys
import _pickle as pickle
import numpy as np
import tensorflow as tf
from keras.preprocessing import image
from keras.applications.vgg16 import VGG16, preprocess_input
from keras.backend.tensorflow_backend import set_session
from keras.models import load_model

os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'

def load_image(path):
    x = preprocess_input(np.expand_dims(image.img_to_array(image.load_img(path, target_size=(224,224,3))), axis=0))
    return np.asarray(x)

def get_model():
	model = VGG16(weights='imagenet', include_top=True, input_shape = (224, 224, 3))
	model.layers.pop()
	model.outputs = [model.layers[-1].output]
	model.layers[-1].outbound_nodes = []
	return model

def get_encoding(model, path):
	image = load_image(path)
	pred = model.predict(image)
	pred = np.reshape(pred, pred.shape[1])
	return pred

embeddings_index = {}
f = open(os.getcwd()+'/files/GLOVE/glove.6B.50d.txt')
for line in f:
    values = line.split()
    word = values[0]
    coefs = np.asarray(values[1:], dtype='float32')
    embeddings_index[word] = coefs
f.close()

path = sys.argv[1]
encoding_model = get_model()
enc = get_encoding(encoding_model,path)
word_index = pickle.load(open('word_pickle.p','rb'))
word_freq = pickle.load(open('word_freq.p','rb'))
rev_word_index = {v: k for k, v in word_index.items()}

# emb = np.zeros(50)


model = load_model(os.getcwd()+'/files/models/learn_embed.hd5')
enc = np.reshape(enc,(1,enc.shape[0]))
emb = np.zeros((20, 7706)); emb[0][word_index['#']] = 1;
emb = np.expand_dims(emb, axis=0)
print (emb.shape)
M = model.predict([enc,emb])

for i in range (1,20):
	a = np.argmax(M[0][i])
	print (rev_word_index[a],)