app.py

import pandas as pd
from huggingface_hub import hf_hub_download
from fastai.learner import load_learner
from flask import Flask, request
import gradio as gr

# Charger le modèle et les données
modelname = 'model_gen0.pkl'
pokemon_types = pd.read_csv('pokemon.csv')
pokemon_types_en = pokemon_types['en']
examplespath = 'images/'
learn_inf = load_learner(hf_hub_download("Okkoman/PokeFace", modelname))

# Créer l'application Flask
app = Flask(__name__)

# Fonction de détection de la langue préférée du client
def detect_language():
    accept_language = request.headers.get("Accept-Language")
    if accept_language:
        languages = [lang.split(";")[0] for lang in accept_language.split(",")]
        return languages[0]
    return 'en'  # Par défaut, en anglais

# Route principale de l'application
@app.route("/")
def index():
    lang = detect_language()
    
    # Définir le titre, la description et le libellé "inconnu" en fonction de la langue
    if lang == 'fr':
        title = "# PokeFace - Quel est ce pokemon ?"
        description = "## Un classifieur pour les pokemons de 1ere et 2eme générations (001-251)"
        unknown = 'inconnu'
    else:
        title = "# PokeFace - What is this pokemon ?"
        description = "## An classifier for 1st-2nd generation pokemons (001-251)"
        unknown = 'unknown'

    # Fonction pour classifier l'image
    def classify_image(img):
        prob_threshold = 0.75
        pred, pred_idx, probs = learn_inf.predict(img)
        index = pokemon_types_en[pokemon_types_en == pred].index[0]
        label = pokemon_types[lang].iloc[index]
        if probs[pred_idx] > prob_threshold:
            return f"{index+1} - {label} ({probs[pred_idx]*100:.0f}%)"
        else:
            return unknown

    # Interface Gradio pour la classification d'image
    with gr.Blocks() as demo:
        with gr.Row():
            gr.Markdown(title)
        with gr.Row():
            gr.Markdown(description)        
        with gr.Row():
            interf = gr.Interface(
                fn=classify_image, 
                inputs=gr.inputs.Image(shape=(192,192)), 
                outputs=gr.outputs.Label(), 
                examples=examplespath, 
                allow_flagging='auto')

    return demo.launch(inline=False)

# Point d'entrée principal
if __name__ == "__main__":
    app.run()