import gradio as gr
import tensorflow as tf
from PIL import Image
import numpy as np


# Klassennamen, sollten Ihrem Dataset entsprechen
class_names = ['Apple__black_rot', 'Apple__healthy', 'Apple__rust', 'Apple__scab']

def predict_figure(uploaded_file):
    if uploaded_file is None:
        return "No file uploaded.", None, "No prediction"

    model = tf.keras.models.load_model('apple_classification.keras')

    # Load the image from the file path
    with Image.open(uploaded_file).convert('RGB') as img:
        img = img.resize((150, 150))
        img_array = np.array(img)

        prediction = model.predict(np.expand_dims(img_array, axis=0))

        # Identify the most confident prediction
        confidences = {class_names[i]: np.round(float(prediction[0][i]), 2) for i in range(len(class_names))}
        return img, confidences

# Define example images
examples = [
    ["images/Apple__black_rot.JPG"],
    ["images/Apple__healthy.JPG"],
    ["images/Apple__rust.JPG"],
    ["images/Apple__scab.JPG"]
]


# Define the Gradio interface
iface = gr.Interface(
    fn=predict_figure,  # Function to process the input
    inputs=gr.File(label="Upload File"),  # File upload widget
    outputs=["image", "text"],  # Output types for image and text
    title="Leaf Disease Classifier",  # Title of the interface
    description="Upload a picture of a Apple Leaf to see condition health it is and the model's confidence level.",  # Description of the interface
    examples=examples  # Example images
)

iface.launch()