Agregando los notebooks exploratorios y el predictor-dl-outputs.ipynb…

… de la última ejecución del model de deep learning
EdwinFLopez · Jun 18, 2024 · e30035c · e30035c
1 parent 6c1caa0
commit e30035c
Show file tree

Hide file tree

Showing 6 changed files with 14,787 additions and 65 deletions.
diff --git a/README.md b/README.md
@@ -7,7 +7,7 @@ Práctica de minería de datos sobre 3 tipos de algoritmos de predicción, usand
 
 * Machine learning con **SVC**: en script `predictor-ml.py`
 * Ensemble con **RandomForest**: en script `predictor-en.py`
-* DeepLearning con **DNN**: en script `predictor-dl.py`
+* DeepLearning con **DNN**: en notebooks `predictor-dl.ipynb` y con la última ejecución: `predictor-dl-outputs.ipynb`  
 
 El objetivo de la predicción es evaluar si un canto de ave pertenece a alguna de las especies observadas
 en el entrenamiento.
@@ -112,9 +112,8 @@ Este archivo se usa para evitar el reprocesamiento de los audios para extraer lo
 
 #### DeepLearning con **DNN**
 
-```shell
-$ python predictor-dl.py
-```
+En este caso, se utiliza un notebook jupyter. El notebook es `predictor-dl.ipynb`. En nuestro caso
+utilizamos tanto PyCharm como VSCode para procesarlo y ejecutarlo.
+El modelo entrenado se almacena en la ruta: `./data/DL_model_birds.h5`.
 
-La primera vez que se ejecuta, crea un archivo del modelo `./data/birds_data_dcnn.csv`.
-Este archivo se usa para evitar el reprocesamiento de los espectogramas.
+Para ver los últimos cálculos generados, se debe cargar el notebook `predictor-dl-outputs.ipynb`. 
diff --git a/features_ext_svm.ipynb b/features_ext_svm.ipynb
diff --git a/predictor-dl-outputs.ipynb b/predictor-dl-outputs.ipynb
diff --git a/predictor-dl.ipynb b/predictor-dl.ipynb
@@ -3,7 +3,7 @@
   {
    "metadata": {},
    "cell_type": "markdown",
-   "source": "# Análisis Espectral de Sonidos de Pájaros Europeos"
+   "source": "# Clasificación de vocalizaciones de aves"
   },
   {
    "metadata": {},
@@ -65,7 +65,7 @@
   {
    "cell_type": "markdown",
    "metadata": {},
-   "source": "## Counting number of spectograms for each bird name"
+   "source": "## Conteo de espectrogramas por ave"
   },
   {
    "cell_type": "code",
@@ -278,46 +278,38 @@
    "source": [
     "from keras.src.losses import SparseCategoricalCrossentropy\n",
     "\n",
-    "# CNN model\n",
-    "model = models.Sequential([\n",
-    "    layers.Conv2D(32, (3, 3), activation='relu', input_shape=(224, 224, 1)),\n",
-    "    layers.MaxPooling2D((2, 2)),\n",
-    "    layers.Conv2D(64, (3, 3), activation='relu'),\n",
-    "    layers.MaxPooling2D((2, 2)),\n",
-    "    layers.Conv2D(128, (3, 3), activation='relu'),\n",
-    "    layers.MaxPooling2D((2, 2)),\n",
-    "    layers.Conv2D(256, (3, 3), activation='relu'),\n",
-    "    layers.Dropout(0.5),\n",
-    "    layers.Flatten(),\n",
-    "    layers.Dense(64, activation='relu'),\n",
-    "    layers.Dropout(0.5),\n",
-    "    layers.Dense(20)    \n",
-    "])\n",
-    "\n",
-    "# Compile the model\n",
-    "model.compile(optimizer='adam', loss=SparseCategoricalCrossentropy(from_logits=True), metrics=['accuracy'])\n",
-    "\n",
-    "# Train the model\n",
-    "model.fit(X_train, y_train, epochs=35)\n"
-   ],
-   "outputs": [],
-   "execution_count": null
-  },
-  {
-   "cell_type": "code",
-   "metadata": {},
-   "source": "model.save(os.path.abspath(\"./data/DL_model_birds.h5\"))",
-   "outputs": [],
-   "execution_count": null
-  },
-  {
-   "cell_type": "code",
-   "metadata": {},
-   "source": [
-    "from tensorflow.keras.models import load_model \n",
-    "\n",
-    "# Load the model\n",
-    "loaded_model = load_model(os.path.abspath(\"./data/DL_model_birds.h5\"))"
+    "# Si el modelo no existe, entrenarlo y guardarlo,\n",
+    "if not os.path.exists(os.path.abspath(\"./data/DL_model_birds.h5\")):\n",
+    "    # CNN model\n",
+    "    model = models.Sequential([\n",
+    "        layers.Conv2D(32, (3, 3), activation='relu', input_shape=(224, 224, 1)),\n",
+    "        layers.MaxPooling2D((2, 2)),\n",
+    "        layers.Conv2D(64, (3, 3), activation='relu'),\n",
+    "        layers.MaxPooling2D((2, 2)),\n",
+    "        layers.Conv2D(128, (3, 3), activation='relu'),\n",
+    "        layers.MaxPooling2D((2, 2)),\n",
+    "        layers.Conv2D(256, (3, 3), activation='relu'),\n",
+    "        layers.Dropout(0.5),\n",
+    "        layers.Flatten(),\n",
+    "        layers.Dense(64, activation='relu'),\n",
+    "        layers.Dropout(0.5),\n",
+    "        layers.Dense(20)    \n",
+    "    ])\n",
+    "    \n",
+    "    # Compilar el modelo\n",
+    "    model.compile(optimizer='adam', loss=SparseCategoricalCrossentropy(from_logits=True), metrics=['accuracy'])\n",
+    "    \n",
+    "    # Entrenar el modelo\n",
+    "    model.fit(X_train, y_train, epochs=35)\n",
+    "    \n",
+    "    # Guardar el modelo\n",
+    "    model.save(os.path.abspath(\"./data/DL_model_birds.h5\"))\n",
+    "    print(f\"Modelo guardado exitosamente en {os.path.abspath('./data/DL_model_birds.h5')}\")\n",
+    "else:\n",
+    "    from tensorflow.keras.models import load_model\n",
+    "    # cargar el modelo\n",
+    "    model = load_model(os.path.abspath(\"./data/DL_model_birds.h5\"))\n",
+    "    print(f\"Cargamos el modelo desde el archivo {os.path.abspath('./data/DL_model_birds.h5')}\")"
    ],
    "outputs": [],
    "execution_count": null
@@ -458,18 +450,10 @@
    "metadata": {},
    "cell_type": "code",
    "source": [
+    "from utils import CLASS_NAMES\n",
+    "\n",
     "# Confusion matrix\n",
     "cm = confusion_matrix(y_test, y_pred)\n",
-    "\n",
-    "# Assuming 'class_names' is a list containing your class labels\n",
-    "# class_names = [\"Acrocephalus arundinaceus\", \"Acrocephalus melanopogon\", \"Acrocephalus scirpaceus\",\n",
-    "#                \"Alcedo atthis\", \"Anas platyrhynchos\", \"Anas strepera\", \"Ardea purpurea\",\n",
-    "#                \"Botaurus stellaris\", \"Charadrius alexandrinus\", \"Ciconia ciconia\",\n",
-    "#                \"Circus aeruginosus\", \"Coracias garrulus\", \"Dendrocopos minor\",\n",
-    "#                \"Fulica atra\", \"Gallinula chloropus\", \"Himantopus himantopus\",\n",
-    "#                \"Ixobrychus minutus\", \"Motacilla flava\", \"Porphyrio porphyrio\", \"Tachybaptus ruficollis\"]\n",
-    "\n",
-    "from utils import CLASS_NAMES\n",
     "conf_matrix_df = pd.DataFrame(cm, index=CLASS_NAMES, columns=CLASS_NAMES)\n",
     "\n",
     "# Plot\n",
@@ -482,13 +466,6 @@
    ],
    "outputs": [],
    "execution_count": null
-  },
-  {
-   "metadata": {},
-   "cell_type": "code",
-   "source": "",
-   "outputs": [],
-   "execution_count": null
   }
  ],
  "metadata": {

diff --git a/reading.ipynb b/reading.ipynb