app.py

import gradio as gr
import torch
from transformers import AutoModelForCausalLM, AutoProcessor
from PIL import Image
import pandas as pd
import numpy as np

# Carrega o modelo Emu3-Gen
model = AutoModelForCausalLM.from_pretrained("BAAI/Emu3-Gen", trust_remote_code=True)
processor = AutoProcessor.from_pretrained("BAAI/Emu3-Gen", trust_remote_code=True)

# Base de dados nutricional
NUTRITION_DB = {
    "arroz": {"calorias": 130, "proteinas": 2.7, "carboidratos": 28, "gorduras": 0.3},
    "feijão": {"calorias": 77, "proteinas": 5.2, "carboidratos": 13.6, "gorduras": 0.5},
    "carne": {"calorias": 250, "proteinas": 26, "carboidratos": 0, "gorduras": 17},
    "batata frita": {"calorias": 312, "proteinas": 3.4, "carboidratos": 41, "gorduras": 15},
    "salada": {"calorias": 15, "proteinas": 1.4, "carboidratos": 2.9, "gorduras": 0.2}
}

def process_image(image, progress=gr.Progress()):
    """Processa a imagem usando Emu3-Gen"""
    try:
        progress(0.3, desc="Processando imagem...")
        
        # Prepara a imagem
        if isinstance(image, str):
            image = Image.open(image)
        elif isinstance(image, np.ndarray):
            image = Image.fromarray(image)

        # Prompt específico para análise de alimentos
        prompt = "Describe in detail all the foods present in this image, including their preparation methods."
        
        progress(0.6, desc="Analisando conteúdo...")
        # Processa a imagem
        inputs = processor(
            images=image,
            text=prompt,
            return_tensors="pt"
        ).to(model.device)
        
        # Gera a descrição
        outputs = model.generate(
            **inputs,
            max_new_tokens=100,
            num_beams=5,
            temperature=0.7,
            top_p=0.9
        )
        
        description = processor.decode(outputs[0], skip_special_tokens=True)
        
        progress(1.0, desc="Concluído!")
        return description
        
    except Exception as e:
        raise gr.Error(f"Erro no processamento: {str(e)}")

def analyze_nutrition(description):
    """Analisa nutrientes baseado na descrição"""
    found_foods = []
    
    # Identifica alimentos na descrição
    for food in NUTRITION_DB.keys():
        if food.lower() in description.lower():
            found_foods.append(food)
    
    # Calcula nutrientes
    total_nutrients = {
        "calorias": 0,
        "proteinas": 0,
        "carboidratos": 0,
        "gorduras": 0
    }
    
    for food in found_foods:
        for nutrient, value in NUTRITION_DB[food].items():
            total_nutrients[nutrient] += value
    
    return total_nutrients, found_foods

def analyze_image(image):
    """Função principal de análise"""
    try:
        # Processa a imagem
        description = process_image(image)
        
        # Analisa nutrientes
        total_nutrients, found_foods = analyze_nutrition(description)
        
        # Prepara dados para visualização
        table_data = [
            ["Calorias", f"{total_nutrients['calorias']:.1f} kcal"],
            ["Proteínas", f"{total_nutrients['proteinas']:.1f}g"],
            ["Carboidratos", f"{total_nutrients['carboidratos']:.1f}g"],
            ["Gorduras", f"{total_nutrients['gorduras']:.1f}g"]
        ]
        
        # Dados para o gráfico
        plot_data = pd.DataFrame({
            'Nutriente': ['Proteínas', 'Carboidratos', 'Gorduras'],
            'Quantidade': [
                total_nutrients['proteinas'],
                total_nutrients['carboidratos'],
                total_nutrients['gorduras']
            ]
        })
        
        # Monta o relatório
        analysis = f"""### 🔍 Análise da Imagem:
{description}

### 🍽️ Alimentos Identificados:
{', '.join(found_foods)}

### 📊 Informação Nutricional:
• Calorias: {total_nutrients['calorias']:.1f} kcal
• Proteínas: {total_nutrients['proteinas']:.1f}g
• Carboidratos: {total_nutrients['carboidratos']:.1f}g
• Gorduras: {total_nutrients['gorduras']:.1f}g

### 💡 Recomendações:
{"⚠️ Alto valor calórico" if total_nutrients['calorias'] > 800 else "✅ Calorias adequadas"}
{"⚠️ Considere reduzir carboidratos" if total_nutrients['carboidratos'] > 60 else "✅ Carboidratos adequados"}
{"⚠️ Alto teor de gorduras" if total_nutrients['gorduras'] > 20 else "✅ Gorduras adequadas"}
"""
        
        return analysis, table_data, plot_data
        
    except Exception as e:
        return str(e), None, None

# Interface Gradio
with gr.Blocks(theme=gr.themes.Soft()) as iface:
    gr.Markdown("# 🍽️ NutriAI - Análise Nutricional com IA")
    
    with gr.Row():
        with gr.Column():
            image_input = gr.Image(
                type="pil",
                label="Foto do Prato",
                sources=["upload", "webcam"]
            )
            analyze_btn = gr.Button("📊 Analisar", variant="primary")
        
        with gr.Column():
            output_text = gr.Markdown()
            
            with gr.Row():
                output_table = gr.Dataframe(
                    headers=["Nutriente", "Quantidade"],
                    label="Informação Nutricional"
                )
                
                output_plot = gr.BarPlot(
                    x="Nutriente",
                    y="Quantidade",
                    title="Macronutrientes (g)",
                    height=300
                )
    
    analyze_btn.click(
        fn=analyze_image,
        inputs=[image_input],
        outputs=[output_text, output_table, output_plot]
    )

if __name__ == "__main__":
    print(f"Usando dispositivo: {'CUDA' if torch.cuda.is_available() else 'CPU'}")
    iface.launch(share=False)