app.py

import numpy as np
import streamlit as st
from tensorflow.keras.preprocessing.sequence import pad_sequences
from tensorflow.keras.models import load_model, Model
from tensorflow.keras.applications.vgg16 import VGG16, preprocess_input
from tensorflow.keras.preprocessing.text import tokenizer_from_json
from tensorflow.keras.preprocessing.image import load_img, img_to_array
from PIL import Image


@st.cache_resource
def init_lstm_model():
    return load_model("./best_model.h5")

@st.cache_resource
def init_vgg16_model():
    vgg_model = VGG16()
    return Model(inputs = vgg_model.inputs , outputs = vgg_model.layers[-2].output)

@st.cache_resource
def init_lstm_tokenizer():
    with open("./tokenizer.txt") as rf:
        return tokenizer_from_json(rf.read())



vgg16_model = init_vgg16_model()
lstm_model = init_lstm_model()
lstm_tokenizer = init_lstm_tokenizer()
max_length = 34

def idx_to_word(integer):
    for word, index in lstm_tokenizer.word_index.items():
        if index == integer:
            return word
    return None


def predict_caption(image, max_length):
    # add start tag for generation process
    in_text = 'startseq'
    # iterate over the max length of sequence
    for _ in range(max_length):
        # encode input sequence
        sequence = lstm_tokenizer.texts_to_sequences([in_text])[0]
        # pad the sequence
        sequence = pad_sequences([sequence], max_length)
        # predict next word
        yhat = lstm_model.predict([image, sequence], verbose=0)
        # get index with high probability
        yhat = np.argmax(yhat)
        # convert index to word
        word = idx_to_word(yhat, lstm_tokenizer)
        # stop if word not found
        if word is None:
            break
        # append word as input for generating next word
        in_text += " " + word
        # stop if we reach end tag
        if word == 'endseq':
            break
    return in_text



def generate_caption(image_name):
    # load the image
    image = load_img(image_name, target_size=(224, 224))
    # convert image pixels to numpy array
    image = img_to_array(image)
    # reshape data for model
    image = image.reshape((1, image.shape[0], image.shape[1], image.shape[2]))
    # preprocess image for vgg
    image = preprocess_input(image)
    feature = vgg16_model.predict(image)
    # predict the caption
    y_pred = predict_caption(feature, max_length)
    return y_pred.repalce("startseq", "").replace("endseq", "").strip()


st.title("""
Image Captioner.
         
This app generates a caption for the input image. The results will be predicted from the basic cnn-rnn to advanced transformer based encoder-decoder models.""")


file_name = st.file_uploader("Upload an image to generate caption...")

if file_name is not None:
    col1, col2 = st.columns(2)

    image = Image.open(file_name)
    col1.image(image, use_column_width=True)
    prediction = generate_caption(file_name)

    col2.header("Predictions")
    col2.subheader(f"VGG16-LSTM : {prediction}")