import gradio as gr
from PIL import Image
import numpy as np
from tensorflow.keras.preprocessing import image as keras_image
from tensorflow.keras.applications.resnet50 import preprocess_input as resnet_preprocess_input
from tensorflow.keras.applications.mobilenet_v2 import preprocess_input as mobilenet_preprocess_input
from tensorflow.keras.models import load_model

# Load your trained models
resnet_model = load_model('/home/user/app/resnet_best_model.keras')  # Update path
mobilenet_model = load_model('/home/user/app/mobilenet_best_model.keras')  # Update path

def predict_comic_character(img, model_type):
    img = Image.fromarray(img.astype('uint8'), 'RGB')
    img = img.resize((224, 224))  # Resize the image to fit model input
    img_array = keras_image.img_to_array(img)
    img_array = np.expand_dims(img_array, axis=0)

    if model_type == 'ResNet50':
        img_array = resnet_preprocess_input(img_array)
        prediction = resnet_model.predict(img_array)
    elif model_type == 'MobileNetV2':
        img_array = mobilenet_preprocess_input(img_array)
        prediction = mobilenet_model.predict(img_array)
    else:
        return {"error": "Invalid model type selected"}

    classes = ['Superman', 'Batman', 'WonderWoman', 'Riddler', 'Spider-Man', 'Iron-Man', 
               'Hulk', 'The Joker', 'Magneto', 'Wolverine', 'Deadpool', 'Catwoman']
    
    return {classes[i]: float(prediction[0][i]) for i in range(len(classes))}

# Define the Gradio interface
interface = gr.Interface(
    fn=predict_comic_character,
    inputs="image",
    outputs="label",
    title="Comic Character Classifier",
    description="Upload an image of a comic character and the classifier will predict the character.",
)

# Launch the interface
interface.launch()