use_model.py

import numpy as np
import sys

from keras import backend as K
def custom_sparse_categorical_accuracy(y_true, y_pred):
    return K.cast(K.equal(K.max(y_true, axis=-1),
                          K.cast(K.argmax(y_pred, axis=-1), K.floatx())),
                  K.floatx())

def input_preprocess(word, tokenizer, sequence_length):
    preprocess_x = tokenizer.texts_to_sequences(word)
    result = [0 for i in range(sequence_length)]
    k = 0
    for i in range(len(preprocess_x)):
        try :
            result[k] = preprocess_x[i][0]
            k = k + 1
        except :
            pass
    return np.array([result]).astype('int32')

def logits_to_text(logits, tokenizer):
    index_to_words = {id: word for word, id in tokenizer.word_index.items()}
    index_to_words[0] = ' '
    return ' '.join([index_to_words[prediction] for prediction in np.argmax(logits, 1)])

def pre_processed_bengali_word(bengali_word):
    result = ""
    for i in bengali_word:
        result = result + i + " "
    return result

def load_files():
    from keras.models import load_model
    model = load_model('transliteration_model.h5', custom_objects={'custom_sparse_categorical_accuracy': custom_sparse_categorical_accuracy})

    import pickle
    with open("ben_token.pickle", "rb") as fp:
        bengali_tokenizer = pickle.load(fp)
    with open("eng_token.pickle", "rb") as fp:
        english_tokenizer = pickle.load(fp)
    with open("max_seq_length.txt", "rb") as fp:
        max_bengali_seq_length = int(fp.read())
    return model, bengali_tokenizer, english_tokenizer, max_bengali_seq_length

def get_prediction(model, bengali_word, bengali_tokenizer, max_seq_length, english_tokenizer):
    p_word = pre_processed_bengali_word(bengali_word)
    preproc_input = input_preprocess(p_word, bengali_tokenizer, max_seq_length)
    prediction = logits_to_text(model.predict(preproc_input)[0], english_tokenizer)
    result = ""
    for i in prediction.strip():
        if i != ' ':
            result = result + i.upper()
    return result


def main():
    usage = "Correct Usage : python3 use_model.py -word <bengali_word>"
    bengali_word = None
    for i, word in enumerate(sys.argv):
        if word == "-word":
            try :
                bengali_word = sys.argv[i + 1]
            except :
                bengali_word == None

    if bengali_word == None:
        print(usage)
        return -1
    model, bengali_tokenizer, english_tokenizer, max_bengali_seq_length = load_files()
    result = get_prediction(model, bengali_word, bengali_tokenizer, max_bengali_seq_length, english_tokenizer)
    with open("output.txt", "w+") as fp:
        fp.write(result)

if __name__ == "__main__":
    main()