catsanddogs_classifier.py

import cv2
import tensorflow as tf

CATEGORIES = ["Dog", "Cat"]

def prepare(filepath):
    IMG_SIZE = 50
    img_array = cv2.imread(filepath, cv2.IMREAD_GRAYSCALE)
    new_array = cv2.resize(img_array, (IMG_SIZE, IMG_SIZE))
    return new_array.reshape(-1, IMG_SIZE, IMG_SIZE,1)
    
#using the saved model directly
model = tf.keras.models.load_model("64x3-CNN.model")

#always predicts using a list, hence []
prediction = model.predict([prepare('dog.jpeg')])
print(CATEGORIES[int(prediction[0][0])])