Update app.py
Browse files
app.py
CHANGED
|
@@ -1,36 +1,32 @@
|
|
| 1 |
-
import os
|
| 2 |
import gradio as gr
|
| 3 |
import torch
|
| 4 |
-
from PIL import Image
|
| 5 |
from transformers import AutoProcessor, AutoModelForVision2Seq
|
| 6 |
-
|
|
|
|
| 7 |
from io import BytesIO
|
| 8 |
|
| 9 |
-
#
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
|
| 24 |
-
|
| 25 |
-
|
| 26 |
-
|
| 27 |
-
|
| 28 |
-
|
| 29 |
-
print(f"⚠️ No se pudo cargar localmente: {e}")
|
| 30 |
-
print("➡️ Se usará la API de Hugging Face.")
|
| 31 |
-
|
| 32 |
-
# ---------- Utilidades ----------
|
| 33 |
def extract_macros(text):
|
|
|
|
| 34 |
def find_value(keyword):
|
| 35 |
m = re.search(rf"{keyword}[^0-9]*([0-9]+)", text.lower())
|
| 36 |
return int(m.group(1)) if m else 0
|
|
@@ -38,7 +34,9 @@ def extract_macros(text):
|
|
| 38 |
kcal = p * 4 + c * 4 + f * 9 if any([p, c, f]) else 0
|
| 39 |
return {"protein": p, "carbs": c, "fat": f, "kcal": kcal}
|
| 40 |
|
|
|
|
| 41 |
def build_macro_card(macros):
|
|
|
|
| 42 |
if not any(macros.values()):
|
| 43 |
return "<div class='card'>⚖️ No se pudieron estimar los macros.</div>"
|
| 44 |
|
|
@@ -63,65 +61,25 @@ def build_macro_card(macros):
|
|
| 63 |
</div>
|
| 64 |
"""
|
| 65 |
|
| 66 |
-
# ---------- Lógica principal ----------
|
| 67 |
-
|
| 68 |
-
def analyze_food(image, text_prompt=""):
|
| 69 |
-
if image is None:
|
| 70 |
-
return "<div class='card'>Subí una imagen del plato 🍽️</div>"
|
| 71 |
-
if not text_prompt.strip():
|
| 72 |
-
text_prompt = "Describe esta comida y estima calorías, proteínas, carbohidratos y grasas."
|
| 73 |
|
|
|
|
|
|
|
| 74 |
try:
|
| 75 |
-
|
| 76 |
-
|
| 77 |
-
|
| 78 |
-
answer = processor.decode(out[0], skip_special_tokens=True)
|
| 79 |
-
else:
|
| 80 |
-
buffered = BytesIO()
|
| 81 |
-
image.save(buffered, format="JPEG")
|
| 82 |
-
img_b64 = base64.b64encode(buffered.getvalue()).decode("utf-8")
|
| 83 |
-
headers = {
|
| 84 |
-
"Authorization": f"Bearer {HF_API_KEY}",
|
| 85 |
-
"Content-Type": "application/json"
|
| 86 |
-
}
|
| 87 |
-
data = {
|
| 88 |
-
"inputs": {
|
| 89 |
-
"text": text_prompt,
|
| 90 |
-
"image": f"data:image/jpeg;base64,{img_b64}"
|
| 91 |
-
}
|
| 92 |
-
}
|
| 93 |
-
|
| 94 |
-
print("📡 Enviando request a:", HF_API_URL)
|
| 95 |
-
r = requests.post(HF_API_URL, headers=headers, json=data)
|
| 96 |
-
|
| 97 |
-
print("📩 Status code:", r.status_code)
|
| 98 |
-
print("🧾 Respuesta cruda:", r.text[:500])
|
| 99 |
-
|
| 100 |
-
# si la respuesta viene vacía
|
| 101 |
-
if not r.text.strip():
|
| 102 |
-
return "<div class='card error'>⚠️ El modelo no devolvió respuesta. Espera unos segundos y reintenta.</div>"
|
| 103 |
-
|
| 104 |
-
# si no es JSON válido
|
| 105 |
-
try:
|
| 106 |
-
resp = r.json()
|
| 107 |
-
except Exception as e:
|
| 108 |
-
return f"<div class='card error'>⚠️ Error al decodificar respuesta: {e}<br><pre>{r.text[:300]}</pre></div>"
|
| 109 |
-
|
| 110 |
-
# Hugging Face a veces devuelve dict con 'error' o 'generated_text'
|
| 111 |
-
if isinstance(resp, dict) and "error" in resp:
|
| 112 |
-
return f"<div class='card error'>⚠️ Error del modelo: {resp['error']}</div>"
|
| 113 |
-
|
| 114 |
-
answer = str(resp)
|
| 115 |
|
| 116 |
macros = extract_macros(answer)
|
| 117 |
card = build_macro_card(macros)
|
| 118 |
return f"<div class='desc'>{answer}</div>{card}"
|
| 119 |
|
| 120 |
except Exception as e:
|
| 121 |
-
return f"<div class='card error'>⚠️ Error
|
| 122 |
|
|
|
|
|
|
|
|
|
|
| 123 |
|
| 124 |
-
# ---------- Interfaz ----------
|
| 125 |
def build_interface():
|
| 126 |
with gr.Blocks(css="""
|
| 127 |
/* --- DARK NEON THEME --- */
|
|
@@ -161,18 +119,16 @@ button {
|
|
| 161 |
}
|
| 162 |
button:hover {opacity:0.8;}
|
| 163 |
""") as demo:
|
| 164 |
-
|
| 165 |
-
|
| 166 |
-
|
| 167 |
-
|
| 168 |
-
|
| 169 |
-
)
|
| 170 |
|
| 171 |
with gr.Row():
|
| 172 |
with gr.Column(scale=1):
|
| 173 |
img = gr.Image(label="📸 Imagen del plato", type="pil")
|
| 174 |
-
txt = gr.Textbox(label="💬 Instrucción (opcional)",
|
| 175 |
-
placeholder="Ej: ¿Cuántas calorías tiene este plato?")
|
| 176 |
btn = gr.Button("🔍 Analizar", variant="primary")
|
| 177 |
with gr.Column(scale=1):
|
| 178 |
out = gr.HTML(label="🧠 Resultado")
|
|
@@ -181,6 +137,7 @@ button:hover {opacity:0.8;}
|
|
| 181 |
|
| 182 |
return demo
|
| 183 |
|
|
|
|
| 184 |
if __name__ == "__main__":
|
| 185 |
demo = build_interface()
|
| 186 |
demo.launch()
|
|
|
|
|
|
|
| 1 |
import gradio as gr
|
| 2 |
import torch
|
|
|
|
| 3 |
from transformers import AutoProcessor, AutoModelForVision2Seq
|
| 4 |
+
from PIL import Image
|
| 5 |
+
import re
|
| 6 |
from io import BytesIO
|
| 7 |
|
| 8 |
+
# ============================================
|
| 9 |
+
# 🔮 CONFIGURACIÓN DEL MODELO
|
| 10 |
+
# ============================================
|
| 11 |
+
|
| 12 |
+
MODEL_ID = "lmms-lab/llava-onevision-1.5-8b-instruct"
|
| 13 |
+
|
| 14 |
+
print("⏳ Cargando modelo local con trust_remote_code=True...")
|
| 15 |
+
processor = AutoProcessor.from_pretrained(MODEL_ID, trust_remote_code=True)
|
| 16 |
+
model = AutoModelForVision2Seq.from_pretrained(
|
| 17 |
+
MODEL_ID,
|
| 18 |
+
trust_remote_code=True,
|
| 19 |
+
torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
|
| 20 |
+
device_map="auto"
|
| 21 |
+
)
|
| 22 |
+
print("✅ Modelo cargado correctamente en modo local.")
|
| 23 |
+
|
| 24 |
+
# ============================================
|
| 25 |
+
# 🧠 FUNCIONES DE ANÁLISIS
|
| 26 |
+
# ============================================
|
| 27 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
| 28 |
def extract_macros(text):
|
| 29 |
+
"""Extrae proteínas, carbohidratos y grasas del texto generado."""
|
| 30 |
def find_value(keyword):
|
| 31 |
m = re.search(rf"{keyword}[^0-9]*([0-9]+)", text.lower())
|
| 32 |
return int(m.group(1)) if m else 0
|
|
|
|
| 34 |
kcal = p * 4 + c * 4 + f * 9 if any([p, c, f]) else 0
|
| 35 |
return {"protein": p, "carbs": c, "fat": f, "kcal": kcal}
|
| 36 |
|
| 37 |
+
|
| 38 |
def build_macro_card(macros):
|
| 39 |
+
"""Genera el HTML visual con barras de progreso tipo dashboard."""
|
| 40 |
if not any(macros.values()):
|
| 41 |
return "<div class='card'>⚖️ No se pudieron estimar los macros.</div>"
|
| 42 |
|
|
|
|
| 61 |
</div>
|
| 62 |
"""
|
| 63 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 64 |
|
| 65 |
+
def analyze_food(image, text_prompt="Describe esta comida y estima sus calorías, proteínas, carbohidratos y grasas."):
|
| 66 |
+
"""Procesa la imagen localmente con el modelo y devuelve descripción + macros."""
|
| 67 |
try:
|
| 68 |
+
inputs = processor(text=text_prompt, images=image, return_tensors="pt").to(model.device)
|
| 69 |
+
out = model.generate(**inputs, max_new_tokens=400)
|
| 70 |
+
answer = processor.decode(out[0], skip_special_tokens=True)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 71 |
|
| 72 |
macros = extract_macros(answer)
|
| 73 |
card = build_macro_card(macros)
|
| 74 |
return f"<div class='desc'>{answer}</div>{card}"
|
| 75 |
|
| 76 |
except Exception as e:
|
| 77 |
+
return f"<div class='card error'>⚠️ Error: {e}</div>"
|
| 78 |
|
| 79 |
+
# ============================================
|
| 80 |
+
# 💅 INTERFAZ DE USUARIO (Glass Neon)
|
| 81 |
+
# ============================================
|
| 82 |
|
|
|
|
| 83 |
def build_interface():
|
| 84 |
with gr.Blocks(css="""
|
| 85 |
/* --- DARK NEON THEME --- */
|
|
|
|
| 119 |
}
|
| 120 |
button:hover {opacity:0.8;}
|
| 121 |
""") as demo:
|
| 122 |
+
|
| 123 |
+
gr.Markdown("""
|
| 124 |
+
<h1>💜 NasFit Vision AI</h1>
|
| 125 |
+
<p>Analiza tus comidas con IA y obtené tu ficha nutricional instantánea.</p>
|
| 126 |
+
""")
|
|
|
|
| 127 |
|
| 128 |
with gr.Row():
|
| 129 |
with gr.Column(scale=1):
|
| 130 |
img = gr.Image(label="📸 Imagen del plato", type="pil")
|
| 131 |
+
txt = gr.Textbox(label="💬 Instrucción (opcional)", placeholder="Ej: ¿Cuántas calorías tiene este plato?")
|
|
|
|
| 132 |
btn = gr.Button("🔍 Analizar", variant="primary")
|
| 133 |
with gr.Column(scale=1):
|
| 134 |
out = gr.HTML(label="🧠 Resultado")
|
|
|
|
| 137 |
|
| 138 |
return demo
|
| 139 |
|
| 140 |
+
|
| 141 |
if __name__ == "__main__":
|
| 142 |
demo = build_interface()
|
| 143 |
demo.launch()
|