File size: 1,766 Bytes
528e68c |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 |
import gradio as gr
import tensorflow as tf
import numpy as np
from PIL import Image
model_path = "pokemon_model_loretmar.keras"
model = tf.keras.models.load_model(model_path)
def predict_pokemon(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, ...]
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_abra = prediction[0][0] # Probability for class 'articuno'
p_aerodactyl = prediction[0][1] # Probability for class 'moltres'
p_eevee = prediction[0][2] # Probability for class 'zapdos'
# return {'charmander': p_charmander, 'mewtwo': p_mewtwo, 'squirtle': p_squirtle}
return {'Abra': p_abra, 'Aerodactyl': p_aerodactyl, 'Eevee': p_eevee}
input_image = gr.Image()
iface = gr.Interface(
fn=predict_pokemon,
inputs=input_image,
outputs=gr.Label(),
examples=["images/00000000.png", "images/00000001.png", "images/00000002.png", "images/00000003.png", "images/00000004.png", "images/00000005.jpg"],
#examples=["pokemon\train\Abra\00000000.png", "pokemon\train\Abra\00000001.png.png", "pokemon\train\Dragonite\00000000.png", "pokemon\train\Dragonite\00000001.png", "pokemon\train\Jigglypuff\00000000.png", "pokemon\train\Jigglypuff\00000001.png"],
description="TEST.")
iface.launch() |