app.py

import gradio as gr
from transformers import AutoImageProcessor, AutoModelForObjectDetection
from PIL import Image
import torch

# Cargar modelo
ckpt = 'yainage90/fashion-object-detection'
processor = AutoImageProcessor.from_pretrained(ckpt)
model = AutoModelForObjectDetection.from_pretrained(ckpt)

# Precios promedios manuales
PRECIOS = {
    'top': 5000,
    'bottom': 15000,
    'shoes': 30000,
    'bag': 20000,
    'outer': 25000
}

def procesar_imagen(img):
    inputs = processor(images=[img], return_tensors="pt")
    outputs = model(**inputs)
    results = processor.post_process_object_detection(outputs, threshold=0.4, target_sizes=[[img.height, img.width]])[0]
    total = 0
    detalle = []
    for score, label, box in zip(results["scores"], results["labels"], results["boxes"]):
        prenda = model.config.id2label[label.item()]
        precio = PRECIOS.get(prenda, 0)
        total += precio
        detalle.append(f"{prenda}: ${precio}")
    texto = "\n".join(detalle + [f"\nTotal aprox.: ${total}"])
    return texto

iface = gr.Interface(fn=procesar_imagen,
                     inputs=gr.Image(type="pil"),
                     outputs="textbox",
                     title="Cuánto vale tu outfit?")
iface.launch()