Upload 5 files

Dockerfile

@@ -1,17 +1,27 @@
-# Imagen base ligera
-FROM python:3.10-slim
+# Dockerfile
 
-# Crear directorio de trabajo
+# Usar una imagen base oficial de Python.
+# 'slim' es una versión ligera, ideal para producción.
+FROM python:3.9-slim
+
+# Establecer el directorio de trabajo dentro del contenedor
 WORKDIR /app
 
-# Copiar archivos
-COPY . .
+# Copiar el archivo de requerimientos primero para aprovechar el cache de Docker.
+# Si los requerimientos no cambian, esta capa no se reconstruirá.
+COPY requirements.txt .
 
-# Instalar dependencias
+# Instalar las dependencias de Python
 RUN pip install --no-cache-dir -r requirements.txt
 
-# Exponer puerto (Hugging Face ignora esto, pero bueno tenerlo)
+# Copiar el resto de los archivos de la aplicación al directorio de trabajo.
+# Esto incluye app.py y tu modelo .h5
+COPY . .
+
+# Exponer el puerto en el que se ejecutará la aplicación.
+# Hugging Face Spaces espera que las apps escuchen en el puerto 7860.
 EXPOSE 7860
 
-# Comando de inicio
-CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]
+# El comando para iniciar la aplicación cuando el contenedor se ejecute.
+# --host 0.0.0.0 es crucial para que la app sea accesible desde fuera del contenedor.
+CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]

app.py

@@ -1,48 +1,85 @@
-from fastapi import FastAPI, UploadFile, File
-from fastapi.responses import JSONResponse
-from PIL import Image, ImageFilter, ImageOps
-import numpy as np
-import tensorflow as tf
-import pickle
-import io
-
-# Cargar modelo y codificador
-model = tf.keras.models.load_model("tbc_cnn_model.h5")
+# app.py
 
-with open("tbc_label_encoder.pkl", "rb") as f:
-    label_encoder = pickle.load(f)
+import io
+import numpy as np
+from PIL import Image
+from fastapi import FastAPI, File, UploadFile
+from fastapi.responses import JSONResponse
+from tensorflow.keras.models import load_model
 
-# Configuración de la app
+# Inicializar la aplicación FastAPI
 app = FastAPI(
-    title="Clasificador TBC con CNN",
-    description="Clasificador de placas de rayos-X de tórax (TBC o normal) usando una CNN",
+    title="API de Clasificación de TBC",
+    description="Una API para clasificar radiografías de tórax como 'normal' o 'tbc' usando un modelo CNN.",
     version="1.0"
 )
 
-def preprocess_image(file: UploadFile, size=(64, 64)) -> np.ndarray:
-    img = Image.open(file.file).convert("L")
-    img = img.resize(size, Image.Resampling.LANCZOS)
-    img = ImageOps.equalize(img)
-    img = img.filter(ImageFilter.EDGE_ENHANCE_MORE)
-    img = img.point(lambda x: 255 if x > 127 else 0, 'L')
-    array = np.array(img) / 255.0
-    array = np.expand_dims(array, axis=-1)   # (64, 64, 1)
-    array = np.expand_dims(array, axis=0)    # (1, 64, 64, 1)
-    return array
+# --- Carga del Modelo ---
+# Cargar el modelo .h5 al iniciar la aplicación para que esté listo para las predicciones.
+# Esto es mucho más eficiente que cargarlo en cada solicitud.
+try:
+    model = load_model('modelo_cnn_tbc.h5')
+    print("Modelo cargado exitosamente.")
+except Exception as e:
+    print(f"Error al cargar el modelo: {e}")
+    model = None
+
+# Definir las constantes del modelo
+IMG_HEIGHT = 150
+IMG_WIDTH = 150
+CLASS_NAMES = ["normal", "tbc"]
+
+# --- Endpoints de la API ---
+
+@app.get("/")
+def read_root():
+    """Endpoint raíz para verificar que la API está funcionando."""
+    return {"message": "Bienvenido a la API de Clasificación de TBC. Usa el endpoint /predict/ para hacer una predicción."}
 
 @app.post("/predict/")
 async def predict(file: UploadFile = File(...)):
+    """
+    Endpoint para predecir si una radiografía es 'normal' o 'tbc'.
+    - Acepta un archivo de imagen (JPG, PNG, etc.).
+    - Devuelve la clase predicha y la confianza de la predicción.
+    """
+    if not model:
+        return JSONResponse(status_code=500, content={"error": "El modelo no está cargado. Revisa los logs del servidor."})
+
+    # 1. Leer el contenido del archivo subido en memoria
+    contents = await file.read()
+    
+    # 2. Convertir los bytes en una imagen PIL
     try:
-        image_array = preprocess_image(file)
-        prediction = model.predict(image_array)
-        predicted_index = np.argmax(prediction)
-        predicted_label = label_encoder.inverse_transform([predicted_index])[0]
-        confidence = float(np.max(prediction)) * 100
+        image = Image.open(io.BytesIO(contents)).convert('RGB')
+    except Exception as e:
+        return JSONResponse(status_code=400, content={"error": f"Archivo inválido. No se pudo procesar la imagen: {e}"})
 
-        return JSONResponse({
-            "label": predicted_label,
-            "confidence": round(confidence, 2)
-        })
+    # 3. Preprocesar la imagen para que coincida con la entrada del modelo
+    # Redimensionar la imagen
+    image = image.resize((IMG_WIDTH, IMG_HEIGHT))
+    # Convertir la imagen a un array de numpy
+    img_array = np.array(image)
+    # Normalizar los valores de los píxeles (de 0-255 a 0-1)
+    img_array = img_array / 255.0
+    # Añadir una dimensión de batch (el modelo espera una forma de [1, 150, 150, 3])
+    image_batch = np.expand_dims(img_array, axis=0)
 
-    except Exception as e:
-        return JSONResponse({"error": str(e)}, status_code=500)
+    # 4. Realizar la predicción
+    prediction = model.predict(image_batch)
+    score = prediction[0][0] # La salida de la sigmoide
+
+    # 5. Interpretar el resultado
+    if score > 0.5:
+        predicted_class = CLASS_NAMES[1]  # 'tbc'
+        confidence = score
+    else:
+        predicted_class = CLASS_NAMES[0]  # 'normal'
+        confidence = 1 - score
+
+    # 6. Devolver el resultado en formato JSON
+    return {
+        "filename": file.filename,
+        "prediction": predicted_class,
+        "confidence": float(confidence)
+    }

modelo_cnn_tbc.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d1f2360569a7f3b20f133e0a2c896c3506ecbf98992647e6de86a9a6e0b306c4
+size 228458584

modelo_cnn_tbc.json

@@ -0,0 +1 @@
+{"module": "keras", "class_name": "Sequential", "config": {"name": "sequential", "trainable": true, "dtype": {"module": "keras", "class_name": "DTypePolicy", "config": {"name": "float32"}, "registered_name": null}, "layers": [{"module": "keras.layers", "class_name": "InputLayer", "config": {"batch_shape": [null, 150, 150, 3], "dtype": "float32", "sparse": false, "name": "input_layer"}, "registered_name": null}, {"module": "keras.layers", "class_name": "Conv2D", "config": {"name": "conv2d", "trainable": true, "dtype": {"module": "keras", "class_name": "DTypePolicy", "config": {"name": "float32"}, "registered_name": null}, "filters": 32, "kernel_size": [3, 3], "strides": [1, 1], "padding": "valid", "data_format": "channels_last", "dilation_rate": [1, 1], "groups": 1, "activation": "relu", "use_bias": true, "kernel_initializer": {"module": "keras.initializers", "class_name": "GlorotUniform", "config": {"seed": null}, "registered_name": null}, "bias_initializer": {"module": "keras.initializers", "class_name": "Zeros", "config": {}, "registered_name": null}, "kernel_regularizer": null, "bias_regularizer": null, "activity_regularizer": null, "kernel_constraint": null, "bias_constraint": null}, "registered_name": null, "build_config": {"input_shape": [null, 150, 150, 3]}}, {"module": "keras.layers", "class_name": "MaxPooling2D", "config": {"name": "max_pooling2d", "trainable": true, "dtype": {"module": "keras", "class_name": "DTypePolicy", "config": {"name": "float32"}, "registered_name": null}, "pool_size": [2, 2], "padding": "valid", "strides": [2, 2], "data_format": "channels_last"}, "registered_name": null}, {"module": "keras.layers", "class_name": "Conv2D", "config": {"name": "conv2d_1", "trainable": true, "dtype": {"module": "keras", "class_name": "DTypePolicy", "config": {"name": "float32"}, "registered_name": null}, "filters": 64, "kernel_size": [3, 3], "strides": [1, 1], "padding": "valid", "data_format": "channels_last", "dilation_rate": [1, 1], "groups": 1, "activation": "relu", "use_bias": true, "kernel_initializer": {"module": "keras.initializers", "class_name": "GlorotUniform", "config": {"seed": null}, "registered_name": null}, "bias_initializer": {"module": "keras.initializers", "class_name": "Zeros", "config": {}, "registered_name": null}, "kernel_regularizer": null, "bias_regularizer": null, "activity_regularizer": null, "kernel_constraint": null, "bias_constraint": null}, "registered_name": null, "build_config": {"input_shape": [null, 74, 74, 32]}}, {"module": "keras.layers", "class_name": "MaxPooling2D", "config": {"name": "max_pooling2d_1", "trainable": true, "dtype": {"module": "keras", "class_name": "DTypePolicy", "config": {"name": "float32"}, "registered_name": null}, "pool_size": [2, 2], "padding": "valid", "strides": [2, 2], "data_format": "channels_last"}, "registered_name": null}, {"module": "keras.layers", "class_name": "Conv2D", "config": {"name": "conv2d_2", "trainable": true, "dtype": {"module": "keras", "class_name": "DTypePolicy", "config": {"name": "float32"}, "registered_name": null}, "filters": 128, "kernel_size": [3, 3], "strides": [1, 1], "padding": "valid", "data_format": "channels_last", "dilation_rate": [1, 1], "groups": 1, "activation": "relu", "use_bias": true, "kernel_initializer": {"module": "keras.initializers", "class_name": "GlorotUniform", "config": {"seed": null}, "registered_name": null}, "bias_initializer": {"module": "keras.initializers", "class_name": "Zeros", "config": {}, "registered_name": null}, "kernel_regularizer": null, "bias_regularizer": null, "activity_regularizer": null, "kernel_constraint": null, "bias_constraint": null}, "registered_name": null, "build_config": {"input_shape": [null, 36, 36, 64]}}, {"module": "keras.layers", "class_name": "MaxPooling2D", "config": {"name": "max_pooling2d_2", "trainable": true, "dtype": {"module": "keras", "class_name": "DTypePolicy", "config": {"name": "float32"}, "registered_name": null}, "pool_size": [2, 2], "padding": "valid", "strides": [2, 2], "data_format": "channels_last"}, "registered_name": null}, {"module": "keras.layers", "class_name": "Flatten", "config": {"name": "flatten", "trainable": true, "dtype": {"module": "keras", "class_name": "DTypePolicy", "config": {"name": "float32"}, "registered_name": null}, "data_format": "channels_last"}, "registered_name": null, "build_config": {"input_shape": [null, 17, 17, 128]}}, {"module": "keras.layers", "class_name": "Dense", "config": {"name": "dense", "trainable": true, "dtype": {"module": "keras", "class_name": "DTypePolicy", "config": {"name": "float32"}, "registered_name": null}, "units": 512, "activation": "relu", "use_bias": true, "kernel_initializer": {"module": "keras.initializers", "class_name": "GlorotUniform", "config": {"seed": null}, "registered_name": null}, "bias_initializer": {"module": "keras.initializers", "class_name": "Zeros", "config": {}, "registered_name": null}, "kernel_regularizer": null, "bias_regularizer": null, "kernel_constraint": null, "bias_constraint": null}, "registered_name": null, "build_config": {"input_shape": [null, 36992]}}, {"module": "keras.layers", "class_name": "Dropout", "config": {"name": "dropout", "trainable": true, "dtype": {"module": "keras", "class_name": "DTypePolicy", "config": {"name": "float32"}, "registered_name": null}, "rate": 0.5, "seed": null, "noise_shape": null}, "registered_name": null}, {"module": "keras.layers", "class_name": "Dense", "config": {"name": "dense_1", "trainable": true, "dtype": {"module": "keras", "class_name": "DTypePolicy", "config": {"name": "float32"}, "registered_name": null}, "units": 1, "activation": "sigmoid", "use_bias": true, "kernel_initializer": {"module": "keras.initializers", "class_name": "GlorotUniform", "config": {"seed": null}, "registered_name": null}, "bias_initializer": {"module": "keras.initializers", "class_name": "Zeros", "config": {}, "registered_name": null}, "kernel_regularizer": null, "bias_regularizer": null, "kernel_constraint": null, "bias_constraint": null}, "registered_name": null, "build_config": {"input_shape": [null, 512]}}], "build_input_shape": [null, 150, 150, 3]}, "registered_name": null, "build_config": {"input_shape": [null, 150, 150, 3]}, "compile_config": {"optimizer": {"module": "keras.optimizers", "class_name": "Adam", "config": {"name": "adam", "learning_rate": 0.0010000000474974513, "weight_decay": null, "clipnorm": null, "global_clipnorm": null, "clipvalue": null, "use_ema": false, "ema_momentum": 0.99, "ema_overwrite_frequency": null, "loss_scale_factor": null, "gradient_accumulation_steps": null, "beta_1": 0.9, "beta_2": 0.999, "epsilon": 1e-07, "amsgrad": false}, "registered_name": null}, "loss": "binary_crossentropy", "loss_weights": null, "metrics": ["accuracy"], "weighted_metrics": null, "run_eagerly": false, "steps_per_execution": 1, "jit_compile": false}}

requirements.txt

@@ -1,7 +1,7 @@
+# requirements.txt
 fastapi
 uvicorn
-tensorflow
-pillow
-scikit-learn
+tensorflow-cpu
 numpy
+pillow
 python-multipart