app.py

import os
import cv2
import random
import numpy as np
from glob import glob
from PIL import Image, ImageOps
import matplotlib.pyplot as plt
import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers

def charbonnier_loss(y_true, y_pred):
    return tf.reduce_mean(tf.sqrt(tf.square(y_true - y_pred) + tf.square(1e-3)))


def peak_signal_noise_ratio(y_true, y_pred):
    return tf.image.psnr(y_pred, y_true, max_val=255.0)
    
 
model = tf.keras.models.load_model('./MIRNet.h5', custom_objects={'charbonnier_loss':                   
charbonnier_loss, "peak_signal_noise_ratio":peak_signal_noise_ratio})

def plot_results(images, titles, figure_size=(12, 12)):
    fig = plt.figure(figsize=figure_size)
    for i in range(len(images)):
        fig.add_subplot(1, len(images), i + 1).set_title(titles[i])
        _ = plt.imshow(images[i])
        plt.axis("off")
    plt.show()


def infer(original_image):
    image = keras.preprocessing.image.img_to_array(original_image)
    image = image.astype("float32") / 255.0
    image = np.expand_dims(image, axis=0)
    output = model.predict(image)
    output_image = output[0] * 255.0
    output_image = output_image.clip(0, 255)
    output_image = output_image.reshape(
        (np.shape(output_image)[0], np.shape(output_image)[1], 3)
    )
    output_image = Image.fromarray(np.uint8(output_image))
    original_image = Image.fromarray(np.uint8(original_image))
    return output_image