import os

import cv2

import numpy as np

import matplotlib.pyplot as plt

os.environ["SM_FRAMEWORK"] = "tf.keras"

import segmentation_models as sm

sm.framework()

my_model = "efficientnetb1"

model = sm.Unet(my_model, encoder_weights="imagenet", input_shape=( 256,256, 3), classes=3, activation='sigmoid')

model.load_weights("Submission_segmentation/weights/unet_model_weights.h5")

#test for single image

def visualize_single_image_prediction(model, image_path):

    # Load the image

    original_img = cv2.imread(image_path)

    resized_img = cv2.resize(original_img, (256, 256))

    # Predict the mask

    X = np.expand_dims(resized_img, 0)

    y_pred = model.predict(X)

    _, y_pred_thr = cv2.threshold(y_pred[0, :, :, 0] * 255, 127, 255, cv2.THRESH_BINARY)

    y_pred = (y_pred_thr / 255).astype(int)

    # Resize the predicted mask back to the original size

    y_pred_original = cv2.resize(y_pred.astype(float), (original_img.shape[1], original_img.shape[0]), interpolation=cv2.INTER_LINEAR)

    # Visualize the original image and predicted mask

    plt.figure(figsize=(12, 4))

    plt.subplot(131)

    plt.title("Original Image")

    plt.imshow(cv2.cvtColor(original_img, cv2.COLOR_BGR2RGB))

    plt.axis("off")

    plt.subplot(132)

    plt.title("Predicted Mask")

    plt.imshow(y_pred_original, cmap="gray")

    plt.axis("off")

    plt.show()

#example

visualize_single_image_prediction(model, "img- (10).png")

print("done")