import numpy as np import cv2 from google.colab.patches import cv2_imshow import os import random import tensorflow as tf h,w = 512,512 num_cases = 10 images = [] labels = [] files = os.listdir('./dataset/images/') random.shuffle(files) model = tf.keras.models.load_model('my_model') lowSevere = 1 midSevere = 2 highSevere = 4 for f in files[0:num_cases]: test_img = cv2.imread('./dataset/images/' + f) resized_img = cv2.resize(test_img,(w,h)) resized_img = resized_img/255 cropped_img = np.reshape(resized_img, (1,resized_img.shape[0],resized_img.shape[1],resized_img.shape[2])) test_out = model.predict(cropped_img) test_out = test_out[0,:,:,0]*1000 test_out = np.clip(test_out,0,255) resized_test_out = cv2.resize(test_out,(test_img.shape[1],test_img.shape[0])) resized_test_out = resized_test_out.astype(np.uint16) test_img = test_img.astype(np.uint16) grey = cv2.cvtColor(test_img, cv2.COLOR_BGR2GRAY) for i in range(test_img.shape[0]): for j in range(test_img.shape[1]): if(grey[i,j]>150 & resized_test_out[i,j]>40): test_img[i,j,1]=test_img[i,j,1] + resized_test_out[i,j] resized_test_out[i,j] = lowSevere elif(grey[i,j]<100 & resized_test_out[i,j]>40): test_img[i,j,2]=test_img[i,j,2] + resized_test_out[i,j] resized_test_out[i,j] = highSevere elif(resized_test_out[i,j]>40): test_img[i,j,0]=test_img[i,j,0] + resized_test_out[i,j] resized_test_out[i,j] = midSevere else: resized_test_out[i,j] = 0 M = cv2.moments(resized_test_out) maxMomentArea = resized_test_out.shape[1]*resized_test_out.shape[0]*highSevere print("0th Moment = " , (M["m00"]*100/maxMomentArea), "%") test_img = np.clip(test_img,0,255) test_img = test_img.astype(np.uint8) cv2_imshow(test_img) cv2.waitKey(0)