import gradio as gr
import tensorflow as tf
import numpy as np
from PIL import Image
 
model_path = "xception.keras"
model = tf.keras.models.load_model(model_path)
 
# Define the core prediction function
def predict_tumor(image):
    # Preprocess image
    print(type(image))
    image = Image.fromarray(image.astype('uint8'))  # Convert numpy array to PIL image
    image = image.resize((150, 150)) #resize the image to 28x28 and converts it to gray scale
    image = np.array(image)
    image = np.expand_dims(image, axis=0) # same as image[None, ...]
   
    # Predict
    prediction = model.predict(image)
 
    # No need to apply sigmoid, as the output layer already uses softmax
    # Convert the probabilities to rounded values
    prediction = np.round(prediction, 2)
 
    # Separate the probabilities for each class
    p_non_tumor = prediction[0][0]  # Probability for class 'charmander'
    tumor = prediction[0][1]   # Probability for class 'mewto'
 
    return {'kein tumor': p_non_tumor, 'tumor': tumor}
 
 
# Create the Gradio interface
input_image = gr.Image()
iface = gr.Interface(
    fn=predict_tumor,
    inputs=input_image,
    outputs=gr.Label(),
    examples=["images/1 no.jpeg", "images/3 no.jpg", "images/2 no.jpeg", "images/5 no.jpg", "images/4 no.jpg", "images/Y1.jpg", "images/Y2.jpg", "images/Y7.jpg", "images/Y4.jpg", "images/Y8.jpg"],  
    description="TEST.")
 
iface.launch()