app.py

import streamlit as st
import tensorflow as tf
from PIL import Image
import numpy as np

st.header("Cat/Dog Classfier")
st.write("This application can be used to classify images whether a cat or a dog is depicted on them.")

selector = st.selectbox(
    label = "Picture or camera input?",
    options = ["picture", "camera"],
    index = 1
    )

if selector == "picture":
    data = st.file_uploader(label="Upload image")

elif selector == "camera":
    data = st.camera_input("Take a picture")

try:
    st.image(data)
except:
    st.warning("No image uploaded!")
    st.stop()

if st.button(label = "Classify"):
    data = Image.open(data)
    image = data.resize((224, 224), Image.LANCZOS)
    #st.image(image)

    # from tensorflow.keras.applications.vgg16 import VGG16
    # vgg16_model = VGG16(include_top=False, input_shape=(224, 224, 3))

    image = np.asarray(image)
    print(image.shape)

    data = image.reshape(1, 224,224,3)
    print(data.shape)

    # X_after_vgg = vgg16_model.predict(data)

    from tensorflow.keras.models import load_model
    model = load_model("TransferlearningMNV2.h5")
    proba = model.predict(data)[0][0]
    st.write(proba)
    prediction = np.round(proba)
    #st.write(prediction)
    #proba = model.predict(data)

    if prediction == 1:
        st.subheader(f"It is to {proba} a cat! 🐈")
    else:
        st.subheader(f"It is to {1-proba} a dog! 🐕 ")