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

# Load the model
model_path = "pokemon-predict-model_transferlearning.keras"
model = tf.keras.models.load_model(model_path)

# Define the core prediction function
def predict_pokemon(image):
    # Preprocess image
    image = Image.fromarray(image.astype('uint8'))  # Convert numpy array to PIL image
    image = image.resize((150, 150))  # Resize the image to 150x150
    image = np.array(image)
    image = np.expand_dims(image, axis=0)  # Same as image[None, ...]

    # Predict
    prediction = model.predict(image)

    # Apply softmax to get probabilities for each class
    prediction = tf.nn.softmax(prediction)

    # Create a dictionary with the probabilities for each Pokemon
    pokemon_classes = [
        'Abra', 'Aerodactyl', 'Alakazam', 'Arbok', 'Arcanine', 'Articuno', 'Beedrill', 'Bellsprout',
        'Blastoise', 'Bulbasaur', 'Butterfree', 'Caterpie', 'Chansey', 'Charizard', 'Charmander',
        'Charmeleon', 'Clefable', 'Clefairy', 'Cloyster', 'Cubone', 'Dewgong', 'Diglett', 'Ditto',
        'Dodrio', 'Doduo', 'Dragonair', 'Dragonite', 'Dratini', 'Drowzee', 'Dugtrio', 'Eevee', 'Ekans',
        'Electabuzz', 'Electrode', 'Exeggcute', 'Exeggutor', 'Farfetchd', 'Fearow', 'Flareon', 'Gastly',
        'Gengar', 'Geodude', 'Gloom', 'Golbat', 'Goldeen', 'Golduck', 'Graveler', 'Grimer', 'Growlithe',
        'Gyarados', 'Haunter', 'Hitmonchan', 'Hitmonlee', 'Horsea', 'Hypno', 'Ivysaur', 'Jigglypuff',
        'Jolteon', 'Jynx', 'Kabutops', 'Kadabra', 'Kakuna', 'Kangaskhan', 'Kingler', 'Koffing', 'Lapras',
        'Lickitung', 'Machamp', 'Machoke', 'Machop', 'Magikarp', 'Magmar', 'Magnemite', 'Magneton', 'Mankey',
        'Marowak', 'Meowth', 'Metapod', 'Mew', 'Mewtwo', 'Moltres', 'Mr. Mime', 'MrMime', 'Nidoking', 'Nidoqueen',
        'Nidorina', 'Nidorino', 'Ninetales', 'Oddish', 'Omanyte', 'Omastar', 'Parasect', 'Pidgeot', 'Pidgeotto',
        'Pidgey', 'Pikachu', 'Pinsir', 'Poliwag', 'Poliwhirl', 'Poliwrath', 'Ponyta', 'Porygon', 'Primeape',
        'Psyduck', 'Raichu', 'Rapidash', 'Raticate', 'Rattata', 'Rhydon', 'Rhyhorn', 'Sandshrew', 'Sandslash',
        'Scyther', 'Seadra', 'Seaking', 'Seel', 'Shellder', 'Slowbro', 'Slowpoke', 'Snorlax', 'Spearow', 'Squirtle',
        'Starmie', 'Staryu', 'Tangela', 'Tauros', 'Tentacool', 'Tentacruel', 'Vaporeon', 'Venomoth', 'Venonat',
        'Venusaur', 'Victreebel', 'Vileplume', 'Voltorb', 'Vulpix', 'Wartortle', 'Weedle', 'Weepinbell', 'Weezing',
        'Wigglytuff', 'Zapdos', 'Zubat'
    ]

    probabilities = [np.round(float(prediction[0][i]), 2) for i in range(len(pokemon_classes))]
    pokemon_probabilities = dict(zip(pokemon_classes, probabilities))

    return pokemon_probabilities

# Interface setup
input_image = gr.Image()
iface = gr.Interface(
    fn=predict_pokemon,
    inputs=input_image,
    outputs=gr.Label(),
    description="A simple MLP classification model for image classification using the Pokémon dataset."
)
iface.launch(share=True)