Add files via upload

The final training monitoring val_loss with patience 40 instead of accuracy

notebooks/ManyShotTransferLearning.ipynb

@@ -17,7 +17,7 @@
     "from tensorflow.keras import optimizers\n",
     "from tensorflow.keras import layers\n",
     "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
-    "from tensorflow.keras.callbacks import Callback\n",
+    "from tensorflow.keras.callbacks import Callback, EarlyStopping, ModelCheckpoint\n",
     "from MLD import multi_lens_distortion\n",
     "\n",
     "os.environ['TF_FORCE_GPU_ALLOW_GROWTH'] = 'true' \n",
@@ -175,29 +175,61 @@
    "source": [
     "model.compile(optimizer=optimizers.Adamax(1e-4), loss=\"CategoricalCrossentropy\", metrics=['accuracy'])\n",
     "\n",
-    "# considering you want to monitor accuracy:\n",
+    "# # considering you want to monitor accuracy:\n",
+    "# acc_thresh = 0.95\n",
     "\n",
-    "acc_thresh = 0.95\n",
+    "# class MyCallback(Callback):\n",
+    "#     def on_epoch_end(self, epoch, logs=None):\n",
+    "#         logs = logs or {}\n",
+    "#         val_accuracy = logs.get('val_accuracy')\n",
+    "#         if val_accuracy is not None and val_accuracy > acc_thresh:\n",
+    "#             print(f'\\nEpoch {epoch}: Early stopping as val_accuracy is {val_accuracy}')\n",
+    "#             self.model.stop_training = True\n",
     "\n",
-    "class MyCallback(Callback):\n",
-    "    def on_epoch_end(self, epoch, logs=None):\n",
-    "        logs = logs or {}\n",
-    "        val_accuracy = logs.get('val_accuracy')\n",
-    "        if val_accuracy is not None and val_accuracy > acc_thresh:\n",
-    "            print(f'\\nEpoch {epoch}: Early stopping as val_accuracy is {val_accuracy}')\n",
-    "            self.model.stop_training = True\n",
     "\n",
+    "# history = model.fit(\n",
+    "#     train_generator,\n",
+    "#     steps_per_epoch=len(train_generator),\n",
+    "#     validation_data=validation_generator,\n",
+    "#     validation_steps=len(validation_generator),\n",
+    "#     epochs=200,\n",
+    "#     callbacks=[MyCallback()]\n",
+    "# )\n",
+    "\n",
+    "# model.save('./PWCModel/')\n",
+    "\n",
+    "\n",
+    "# Parameters for EarlyStopping and ModelCheckpoint\n",
+    "patience = 40\n",
+    "\n",
+    "# Setting up callbacks for early stopping on minimum validation loss and saving the best model\n",
+    "early_stopping_callback = EarlyStopping(\n",
+    "    monitor='val_loss',\n",
+    "    patience=patience,\n",
+    "    verbose=1,\n",
+    "    mode='min',\n",
+    "    restore_best_weights=True  # Restores model weights from the epoch with the best value of the monitored quantity.\n",
+    ")\n",
+    "\n",
+    "model_checkpoint_callback = ModelCheckpoint(\n",
+    "    './PWCModel/best_model.h5',  # Path where the model will be saved\n",
+    "    monitor='val_loss',\n",
+    "    save_best_only=True,  # Only the best model according to the validation loss is saved\n",
+    "    mode='min',\n",
+    "    verbose=1\n",
+    ")\n",
     "\n",
     "history = model.fit(\n",
     "    train_generator,\n",
     "    steps_per_epoch=len(train_generator),\n",
     "    validation_data=validation_generator,\n",
     "    validation_steps=len(validation_generator),\n",
     "    epochs=200,\n",
-    "    callbacks=[MyCallback()]\n",
+    "    callbacks=[early_stopping_callback, model_checkpoint_callback]\n",
     ")\n",
     "\n",
-    "model.save('./PWCModel/')"
+    "# Save the overall model after training (optional, as the best model is already saved)\n",
+    "model.save('./PWCModel/best_PWC_model.h5')\n"
    ]
   },
   {