app.py

from PIL import Image
import streamlit as st
from streamlit_drawable_canvas import st_canvas
import tensorflow_addons as tfa
import tensorflow as tf
import numpy as np
from tensorflow.keras.utils import custom_object_scope

# Define a function to create the InstanceNormalization layer
def create_in():
    return tfa.layers.InstanceNormalization()


def model_out(model_path,img):
  with custom_object_scope({'InstanceNormalization': create_in}):
    model = tf.keras.models.load_model(model_path)
  img = (img-127.5)/127.5
  img = np.expand_dims(img, 0)
  pred = model.predict(img)
  pred = np.asarray(pred)
  return pred[0]

# Specify canvas parameters in application
drawing_mode = st.sidebar.selectbox(
    "Drawing tool:", ("freedraw", "line")
)

stroke_width = st.sidebar.slider("Stroke width: ", 1, 25, 3)

realtime_update = st.sidebar.checkbox("Update in realtime", True)
# Create a canvas component
canvas_result = st_canvas(
    stroke_width=stroke_width,
    stroke_color='#000000',
    background_color='#FFFFFF',
    update_streamlit=realtime_update,
    height=256,
    width=256,
    drawing_mode=drawing_mode,
    key="canvas",
)

# Do something interesting with the image data and paths
if canvas_result.image_data is not None:
    st.image(canvas_result.image_data)
    img = np.array(canvas_result.image_data)
    img_rgb = Image.fromarray(img).convert("L")
    img = np.array(img_rgb)
    pred = model_out('Sketch2Car.h5', img)
    pred = (pred + 1.0) / 2.0  # Undo tanh normalization to get values in [0.0, 1.0] range
    pred = Image.fromarray((pred * 255).astype(np.uint8)).convert("RGB")
    st.image(pred)