app.py

#! pip install gradio
import gradio as gr
from transformers import pipeline

# Lade beide Modelle
vit_classifier = pipeline("image-classification", model="chrisis2/vit-food-classification-chrisis2")  # Ersetze mit deinem Modell
clip_detector = pipeline(model="openai/clip-vit-large-patch14", task="zero-shot-image-classification")

# Verwende deine Food-Labels
food_labels = [
    "Baked Potato", "Crispy Chicken", "Donut", "Fries", "Hot Dog", "Sandwich", "Taco", "Taquito",
    "apple_pie", "burger", "butter_naan", "chai", "chapati", "cheesecake", "chicken_curry",
    "chole_bhature", "dal_makhani", "dhokla", "fried_rice", "ice_cream", "idli", "jalebi",
    "kaathi_rolls", "kadai_paneer", "kulfi", "masala_dosa", "momos", "omelette", "paani_puri",
    "pakode", "pav_bhaji", "pizza", "samosa", "sushi"
]

def classify_food(image):
    # Klassifiziere mit deinem trainierten Modell
    vit_results = vit_classifier(image)
    vit_output = {result['label']: result['score'] for result in vit_results}
    
    # Klassifiziere mit CLIP Zero-Shot
    clip_results = clip_detector(image, candidate_labels=food_labels)
    clip_output = {result['label']: result['score'] for result in clip_results}
    
    return {"Trainiertes ViT-Modell": vit_output, "CLIP Zero-Shot": clip_output}

# Füge Beispielbilder aus deinem Dataset hinzu
example_images = [
    ["gradio-food-app/example_images/burger.jpg.jpg"],
    ["gradio-food-app/example_images/burger.jpg.jpg"],
    ["gradio-food-app/example_images/burger.jpg.jpg"],
    ["gradio-food-app/example_images/pizza.jpg.jpg"],
    ["gradio-food-app/example_images/sushi.jpg.jpg"],
    # Füge weitere Beispiele hinzu
]

# Erstelle die Gradio-Oberfläche
iface = gr.Interface(
    fn=classify_food,
    inputs=gr.Image(type="filepath"),
    outputs=gr.JSON(),
    title="Food Classification Vergleich",
    description="Lade ein Bild eines Lebensmittels hoch und vergleiche die Ergebnisse eines trainierten ViT-Modells mit einem Zero-Shot CLIP-Modell.",
    examples=example_images
)

iface.launch()