app.py

import gradio as gr
from transformers import pipeline
from PIL import Image
import torch
from torchvision import transforms

INGREDIENT_MODEL_ID = "stchakman/Fridge_Items_Model"
ingredient_classifier = pipeline(
    "image-classification",
    model=INGREDIENT_MODEL_ID,
    device=0 if torch.cuda.is_available() else -1,
    top_k=5 
)

RECIPE_MODEL_ID = "flax-community/t5-recipe-generation"
recipe_generator = pipeline("text2text-generation", model=RECIPE_MODEL_ID)

augment = transforms.Compose([
    transforms.RandomHorizontalFlip(p=0.5),
    transforms.RandomRotation(10),
    transforms.ColorJitter(brightness=0.2, contrast=0.2),
])

def generate_recipe(image: Image.Image):
    image_aug = augment(image)
    results = ingredient_classifier(image_aug)
    ingredients = [res["label"] for res in results]
    
    ingredient_str = ", ".join(ingredients)

    prompt = f"Ingredients: {ingredient_str}. Recipe:"
    recipe = recipe_generator(prompt, max_length=300, do_sample=True)[0]["generated_text"]
    
    return f"### Ingredients detectes :\n{ingredient_str}\n\n### Recette generee :\n{recipe}"

interface = gr.Interface(
    fn=generate_recipe,
    inputs=gr.Image(type="pil"),
    outputs=gr.Markdown(),
    title="🥕 Generateur de recettes 🧑‍🍳",
    description="Envoyer une image d'ingredients pour recevoir une recette",
    allow_flagging="never"
)

if __name__ == "__main__":
    interface.launch()