predict.py

import numpy as np
import matplotlib.pyplot as plt
from PIL import Image
from tensorflow.keras.models import load_model

model = load_model('my_model.keras')
model.summary()

def predict(img_path):
    class_dict = {'glioma': 0, 'meningioma': 1, 'notumor': 2, 'pituitary': 3}
    
    label = list(class_dict.keys())
    plt.figure(figsize=(12, 12))
    img = Image.open(img_path)
    resized_img = img.resize((299, 299))
    img = np.asarray(resized_img)
    img = np.expand_dims(img, axis=0)
    img = img / 255
    predictions = model.predict(img)
    probs = list(predictions[0])
    labels = label
    plt.subplot(2, 1, 1)
    plt.imshow(resized_img)
    plt.subplot(2, 1, 2)
    bars = plt.barh(labels, probs)
    plt.xlabel('Olasılık', fontsize=15)
    ax = plt.gca()
    ax.bar_label(bars, fmt = '%.2f')
    plt.show()



predict("Testing/notumor/Te-no_0010.jpg")