# AUTOGENERATED! DO NOT EDIT! File to edit: app.ipynb.

# %% auto 0
__all__ = ['path', 'learn_inf', 'image', 'label', 'examples', 'intf', 'on_click_classify']

# %% app.ipynb 2
from fastai.vision.all import *
import gradio as gr
from fastai.vision.widgets import *

# %% app.ipynb 12
path = Path('.')
learn_inf = load_learner(path/'export.pkl')

# %% app.ipynb 14
from PIL import Image
import ipywidgets as widgets

# Optional: Import display only if in an IPython environment
try:
    from IPython.display import display
    can_display = True
except ImportError:
    can_display = False

def on_click_classify(img_array):
    # Convert numpy array to PIL Image
    img = Image.fromarray(img_array.astype('uint8'), 'RGB')
    
    out_pl = widgets.Output()
    out_pl.clear_output()
    if can_display:
        # Use display if available
        with out_pl:
            display(img.to_thumb(128, 128))
    else:
        # Save to a file if display is not available
        img.to_thumb(128, 128).save('output_thumbnail.png')
        print("Thumbnail saved to 'output_thumbnail.png'.")
    
    # Assuming learn_inf is already defined and loaded elsewhere in your code
    pred, pred_idx, probs = learn_inf.predict(img)
    return f'Prediction: {pred}; Probability: {probs[pred_idx]:.04f}'

# %% app.ipynb 17
image = gr.Image()
label = gr.Label()
examples = ['Adi_trainers.jpg', 'Nike_trainers.jpg', 'Puma_trainers.jpg', 'Adidas_trainers.jpg']

intf = gr.Interface(fn=on_click_classify, inputs=image, outputs=label, examples=examples)
intf.launch(inline=False)