import streamlit as st
import cv2
from PIL import Image
import numpy as np
import tensorflow as tf
from tensorflow.keras.applications.resnet50 import preprocess_input 
from tensorflow.keras.preprocessing.image import img_to_array    

st.title('Jacaranda Identification')
st.markdown('A Deep learning application to identify if a satellite image clip contains Jacaranda trees. The predicting result will be "Jacaranda", or "Others".')

uploaded_file = st.file_uploader("Upload an image file", type="jpg")

img_height = 224
img_width = 224
class_names = ['Jacaranda', 'Others']

model = tf.keras.models.load_model('model')

if uploaded_file is not None:
    img = Image.open(uploaded_file)
    st.image(img)

    img_array = img_to_array(img)
    img_array = tf.expand_dims(img_array, axis = 0) # Create a batch
    processed_image = preprocess_input(img_array)

    Generate_pred = st.button("Generate Prediction")
    if Generate_pred:
        predictions = model.predict(processed_image)
        score = predictions[0]
        st.markdown("Predicted class of the image is : {}".format(class_names[np.argmax(score)]))