02_network_training_tf.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-09-29 18:49:03.195430: I tensorflow/core/platform/cpu_feature_guard.cc:193] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  AVX2 AVX_VNNI FMA\n",
      "To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.\n",
      "2022-09-29 18:49:03.398146: I tensorflow/core/util/util.cc:169] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`.\n",
      "2022-09-29 18:49:03.469240: E tensorflow/stream_executor/cuda/cuda_blas.cc:2981] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered\n",
      "2022-09-29 18:49:04.239127: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libnvinfer.so.7'; dlerror: libnvinfer.so.7: cannot open shared object file: No such file or directory\n",
      "2022-09-29 18:49:04.239213: W tensorflow/stream_executor/platform/default/dso_loader.cc:64] Could not load dynamic library 'libnvinfer_plugin.so.7'; dlerror: libnvinfer_plugin.so.7: cannot open shared object file: No such file or directory\n",
      "2022-09-29 18:49:04.239219: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Cannot dlopen some TensorRT libraries. If you would like to use Nvidia GPU with TensorRT, please make sure the missing libraries mentioned above are installed properly.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tf version: 2.10.0\n",
      "GPUs:  [PhysicalDevice(name='/physical_device:GPU:0', device_type='GPU')]\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-09-29 18:49:05.296302: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:966] could not open file to read NUMA node: /sys/bus/pci/devices/0000:01:00.0/numa_node\n",
      "Your kernel may have been built without NUMA support.\n",
      "2022-09-29 18:49:05.350499: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:966] could not open file to read NUMA node: /sys/bus/pci/devices/0000:01:00.0/numa_node\n",
      "Your kernel may have been built without NUMA support.\n",
      "2022-09-29 18:49:05.350532: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:966] could not open file to read NUMA node: /sys/bus/pci/devices/0000:01:00.0/numa_node\n",
      "Your kernel may have been built without NUMA support.\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import numpy as np\n",
    "import pickle\n",
    "import matplotlib.pyplot as plt\n",
    "from datetime import datetime\n",
    "import random\n",
    "\n",
    "import tensorflow as tf\n",
    "from tensorflow.keras.models import Sequential, Model\n",
    "from tensorflow.keras.layers import Dense, Dropout, Activation, Flatten, Lambda, Concatenate, BatchNormalization\n",
    "from tensorflow.keras.layers import Input, Add, Conv2D, MaxPooling2D, AveragePooling2D, Softmax, InputLayer\n",
    "from tensorflow.keras.utils import plot_model\n",
    "from tensorflow.keras.callbacks import ModelCheckpoint\n",
    "from keras import backend as K\n",
    "\n",
    "# set random seed value for reproducable results\n",
    "seed_value= 4815\n",
    "os.environ['PYTHONHASHSEED']=str(seed_value)\n",
    "random.seed(seed_value)\n",
    "np.random.seed(seed_value)\n",
    "tf.random.set_seed(seed_value)\n",
    "\n",
    "print(f\"tf version: {tf.__version__}\")\n",
    "print(\"GPUs: \", tf.config.list_physical_devices('GPU'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "EPOCHS = 50\n",
    "BATCH_SIZE_TRAIN = 64\n",
    "VALIDATION = True"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# load data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "number training data: 78720\n",
      "batch_size_train: 64\n",
      "number validation data: 38528\n",
      "batch_size_val: 64\n"
     ]
    }
   ],
   "source": [
    "data_path = \"data/\"\n",
    "train_path = data_path + \"train2014/train2014\"\n",
    "val_path = data_path + \"val2014/val2014\"\n",
    "\n",
    "with open(os.path.join(train_path +\"_processed.pickle\"), \"rb\") as f:\n",
    "    training_data = pickle.load(f)\n",
    "len_training_data_raw = len(training_data)\n",
    "\n",
    "with open(os.path.join(val_path + \"_processed.pickle\"), \"rb\") as f:\n",
    "    validation_data = pickle.load(f)\n",
    "\n",
    "training_data = training_data[:len(training_data) // BATCH_SIZE_TRAIN * BATCH_SIZE_TRAIN]\n",
    "print(f\"number training data: {len(training_data)}\")\n",
    "print(\"batch_size_train: \" + str(BATCH_SIZE_TRAIN))\n",
    "if VALIDATION:\n",
    "    # batch_size_val = int(BATCH_SIZE_TRAIN * (len(validation_data) / (len_training_data_raw + len(validation_data))))\n",
    "    batch_size_val = BATCH_SIZE_TRAIN\n",
    "    validation_data = validation_data[:len(validation_data) // batch_size_val * batch_size_val]\n",
    "    print(f\"number validation data: {len(validation_data)}\")\n",
    "    print(f\"batch_size_val: {batch_size_val}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# construct network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-09-29 18:49:26.019327: I tensorflow/core/platform/cpu_feature_guard.cc:193] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  AVX2 AVX_VNNI FMA\n",
      "To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.\n",
      "2022-09-29 18:49:26.021199: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:966] could not open file to read NUMA node: /sys/bus/pci/devices/0000:01:00.0/numa_node\n",
      "Your kernel may have been built without NUMA support.\n",
      "2022-09-29 18:49:26.021269: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:966] could not open file to read NUMA node: /sys/bus/pci/devices/0000:01:00.0/numa_node\n",
      "Your kernel may have been built without NUMA support.\n",
      "2022-09-29 18:49:26.021282: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:966] could not open file to read NUMA node: /sys/bus/pci/devices/0000:01:00.0/numa_node\n",
      "Your kernel may have been built without NUMA support.\n",
      "2022-09-29 18:49:26.870337: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:966] could not open file to read NUMA node: /sys/bus/pci/devices/0000:01:00.0/numa_node\n",
      "Your kernel may have been built without NUMA support.\n",
      "2022-09-29 18:49:26.870714: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:966] could not open file to read NUMA node: /sys/bus/pci/devices/0000:01:00.0/numa_node\n",
      "Your kernel may have been built without NUMA support.\n",
      "2022-09-29 18:49:26.870730: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1700] Could not identify NUMA node of platform GPU id 0, defaulting to 0.  Your kernel may not have been built with NUMA support.\n",
      "2022-09-29 18:49:26.870754: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:966] could not open file to read NUMA node: /sys/bus/pci/devices/0000:01:00.0/numa_node\n",
      "Your kernel may have been built without NUMA support.\n",
      "2022-09-29 18:49:26.871010: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1616] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 5440 MB memory:  -> device: 0, name: NVIDIA GeForce RTX 3060 Ti, pci bus id: 0000:01:00.0, compute capability: 8.6\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " conv2d (Conv2D)             (None, 128, 128, 64)      1216      \n",
      "                                                                 \n",
      " batch_normalization (BatchN  (None, 128, 128, 64)     256       \n",
      " ormalization)                                                   \n",
      "                                                                 \n",
      " conv2d_1 (Conv2D)           (None, 128, 128, 64)      36928     \n",
      "                                                                 \n",
      " batch_normalization_1 (Batc  (None, 128, 128, 64)     256       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " max_pooling2d (MaxPooling2D  (None, 64, 64, 64)       0         \n",
      " )                                                               \n",
      "                                                                 \n",
      " conv2d_2 (Conv2D)           (None, 64, 64, 64)        36928     \n",
      "                                                                 \n",
      " batch_normalization_2 (Batc  (None, 64, 64, 64)       256       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " conv2d_3 (Conv2D)           (None, 64, 64, 64)        36928     \n",
      "                                                                 \n",
      " batch_normalization_3 (Batc  (None, 64, 64, 64)       256       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " max_pooling2d_1 (MaxPooling  (None, 32, 32, 64)       0         \n",
      " 2D)                                                             \n",
      "                                                                 \n",
      " conv2d_4 (Conv2D)           (None, 32, 32, 128)       73856     \n",
      "                                                                 \n",
      " batch_normalization_4 (Batc  (None, 32, 32, 128)      512       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " conv2d_5 (Conv2D)           (None, 32, 32, 128)       147584    \n",
      "                                                                 \n",
      " batch_normalization_5 (Batc  (None, 32, 32, 128)      512       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " max_pooling2d_2 (MaxPooling  (None, 16, 16, 128)      0         \n",
      " 2D)                                                             \n",
      "                                                                 \n",
      " conv2d_6 (Conv2D)           (None, 16, 16, 128)       147584    \n",
      "                                                                 \n",
      " batch_normalization_6 (Batc  (None, 16, 16, 128)      512       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " conv2d_7 (Conv2D)           (None, 16, 16, 128)       147584    \n",
      "                                                                 \n",
      " batch_normalization_7 (Batc  (None, 16, 16, 128)      512       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " flatten (Flatten)           (None, 32768)             0         \n",
      "                                                                 \n",
      " dropout (Dropout)           (None, 32768)             0         \n",
      "                                                                 \n",
      " dense (Dense)               (None, 1024)              33555456  \n",
      "                                                                 \n",
      " dropout_1 (Dropout)         (None, 1024)              0         \n",
      "                                                                 \n",
      " dense_1 (Dense)             (None, 8)                 8200      \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 34,195,336\n",
      "Trainable params: 34,193,800\n",
      "Non-trainable params: 1,536\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "def homography_net():\n",
    "    input_shape = (128, 128, 2)\n",
    "    kernel_size = 3\n",
    "    pool_size = 2\n",
    "    channels = 64\n",
    "    dropout = 0.5\n",
    "\n",
    "    model = Sequential()\n",
    "    model.add(InputLayer(input_shape))\n",
    "    model.add(Conv2D(filters=channels,\\\n",
    "            kernel_size=kernel_size, activation='relu', padding='same'))\n",
    "    model.add(BatchNormalization())\n",
    "    model.add(Conv2D(filters=channels,\\\n",
    "            kernel_size=kernel_size, activation='relu', padding='same'))\n",
    "    model.add(BatchNormalization())\n",
    "    model.add(MaxPooling2D(pool_size))\n",
    "\n",
    "    model.add(Conv2D(filters=channels,\\\n",
    "            kernel_size=kernel_size, activation='relu', padding='same'))\n",
    "    model.add(BatchNormalization())\n",
    "    model.add(Conv2D(filters=channels,\\\n",
    "            kernel_size=kernel_size, activation='relu', padding='same'))\n",
    "    model.add(BatchNormalization())\n",
    "    model.add(MaxPooling2D(pool_size))\n",
    "\n",
    "    model.add(Conv2D(filters=channels*2,\\\n",
    "            kernel_size=kernel_size, activation='relu', padding='same',))\n",
    "    model.add(BatchNormalization())\n",
    "    model.add(Conv2D(filters=channels*2,\\\n",
    "            kernel_size=kernel_size, activation='relu', padding='same',))\n",
    "    model.add(BatchNormalization())\n",
    "    model.add(MaxPooling2D(pool_size))\n",
    "\n",
    "    model.add(Conv2D(filters=channels*2,\\\n",
    "            kernel_size=kernel_size, activation='relu', padding='same',))\n",
    "    model.add(BatchNormalization())\n",
    "    model.add(Conv2D(filters=channels*2,\\\n",
    "            kernel_size=kernel_size, activation='relu', padding='same',))\n",
    "    model.add(BatchNormalization())\n",
    "\n",
    "    model.add(Flatten())\n",
    "    model.add(Dropout(dropout))\n",
    "    model.add(Dense(1024, activation='relu'))\n",
    "    model.add(Dropout(dropout))\n",
    "    model.add(Dense(8))\n",
    "    return model\n",
    "\n",
    "model = homography_net()\n",
    "sgd = tf.optimizers.SGD(learning_rate=0.005, momentum=0.9)\n",
    "\n",
    "def euclidean_l2(y_true, y_pred):\n",
    "    return K.sqrt(K.sum(K.square(y_pred - y_true), axis=-1, keepdims=True))\n",
    "model.compile(optimizer=sgd, loss=euclidean_l2, metrics=['mean_squared_error'])\n",
    "\n",
    "print(model.summary())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# train network"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/50\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-09-29 18:49:28.864331: I tensorflow/stream_executor/cuda/cuda_dnn.cc:384] Loaded cuDNN version 8500\n",
      "2022-09-29 18:49:30.184500: I tensorflow/stream_executor/cuda/cuda_blas.cc:1614] TensorFloat-32 will be used for the matrix multiplication. This will only be logged once.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1230/1230 [==============================] - 171s 135ms/step - loss: 0.8591 - mean_squared_error: 0.1879 - val_loss: 0.5591 - val_mean_squared_error: 0.0504\n",
      "Epoch 2/50\n",
      "1230/1230 [==============================] - 164s 133ms/step - loss: 0.6338 - mean_squared_error: 0.0629 - val_loss: 0.4916 - val_mean_squared_error: 0.0386\n",
      "Epoch 3/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.5822 - mean_squared_error: 0.0538 - val_loss: 0.4526 - val_mean_squared_error: 0.0325\n",
      "Epoch 4/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.5335 - mean_squared_error: 0.0460 - val_loss: 0.4066 - val_mean_squared_error: 0.0279\n",
      "Epoch 5/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.4853 - mean_squared_error: 0.0390 - val_loss: 0.3701 - val_mean_squared_error: 0.0241\n",
      "Epoch 6/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.4393 - mean_squared_error: 0.0329 - val_loss: 0.3255 - val_mean_squared_error: 0.0192\n",
      "Epoch 7/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.3991 - mean_squared_error: 0.0281 - val_loss: 0.3035 - val_mean_squared_error: 0.0182\n",
      "Epoch 8/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.3677 - mean_squared_error: 0.0245 - val_loss: 0.3086 - val_mean_squared_error: 0.0191\n",
      "Epoch 9/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.3430 - mean_squared_error: 0.0219 - val_loss: 0.2606 - val_mean_squared_error: 0.0136\n",
      "Epoch 10/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.3207 - mean_squared_error: 0.0198 - val_loss: 0.2464 - val_mean_squared_error: 0.0117\n",
      "Epoch 11/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.3048 - mean_squared_error: 0.0182 - val_loss: 0.2361 - val_mean_squared_error: 0.0115\n",
      "Epoch 12/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.2878 - mean_squared_error: 0.0164 - val_loss: 0.2342 - val_mean_squared_error: 0.0109\n",
      "Epoch 13/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.2777 - mean_squared_error: 0.0155 - val_loss: 0.2195 - val_mean_squared_error: 0.0095\n",
      "Epoch 14/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.2658 - mean_squared_error: 0.0141 - val_loss: 0.2939 - val_mean_squared_error: 0.0177\n",
      "Epoch 15/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.2566 - mean_squared_error: 0.0133 - val_loss: 0.2614 - val_mean_squared_error: 0.0133\n",
      "Epoch 16/50\n",
      "1230/1230 [==============================] - 162s 131ms/step - loss: 0.2471 - mean_squared_error: 0.0124 - val_loss: 0.2147 - val_mean_squared_error: 0.0092\n",
      "Epoch 17/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.2385 - mean_squared_error: 0.0115 - val_loss: 0.1987 - val_mean_squared_error: 0.0081\n",
      "Epoch 18/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.2295 - mean_squared_error: 0.0107 - val_loss: 0.2114 - val_mean_squared_error: 0.0084\n",
      "Epoch 19/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.2237 - mean_squared_error: 0.0102 - val_loss: 0.1778 - val_mean_squared_error: 0.0069\n",
      "Epoch 20/50\n",
      "1230/1230 [==============================] - 162s 131ms/step - loss: 0.2155 - mean_squared_error: 0.0094 - val_loss: 0.1966 - val_mean_squared_error: 0.0080\n",
      "Epoch 21/50\n",
      "1230/1230 [==============================] - 162s 131ms/step - loss: 0.2096 - mean_squared_error: 0.0090 - val_loss: 0.2286 - val_mean_squared_error: 0.0125\n",
      "Epoch 22/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.2041 - mean_squared_error: 0.0085 - val_loss: 0.2045 - val_mean_squared_error: 0.0080\n",
      "Epoch 23/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.1967 - mean_squared_error: 0.0077 - val_loss: 0.1828 - val_mean_squared_error: 0.0075\n",
      "Epoch 24/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.1907 - mean_squared_error: 0.0073 - val_loss: 0.1673 - val_mean_squared_error: 0.0062\n",
      "Epoch 25/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.1856 - mean_squared_error: 0.0069 - val_loss: 0.1864 - val_mean_squared_error: 0.0068\n",
      "Epoch 26/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.1810 - mean_squared_error: 0.0064 - val_loss: 0.1340 - val_mean_squared_error: 0.0046\n",
      "Epoch 27/50\n",
      "1230/1230 [==============================] - 162s 131ms/step - loss: 0.1763 - mean_squared_error: 0.0061 - val_loss: 0.1615 - val_mean_squared_error: 0.0059\n",
      "Epoch 28/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.1733 - mean_squared_error: 0.0059 - val_loss: 0.1330 - val_mean_squared_error: 0.0044\n",
      "Epoch 29/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.1705 - mean_squared_error: 0.0057 - val_loss: 0.1568 - val_mean_squared_error: 0.0055\n",
      "Epoch 30/50\n",
      "1230/1230 [==============================] - 173s 141ms/step - loss: 0.1676 - mean_squared_error: 0.0055 - val_loss: 0.1432 - val_mean_squared_error: 0.0058\n",
      "Epoch 31/50\n",
      "1230/1230 [==============================] - 172s 140ms/step - loss: 0.1641 - mean_squared_error: 0.0053 - val_loss: 0.1317 - val_mean_squared_error: 0.0044\n",
      "Epoch 32/50\n",
      "1230/1230 [==============================] - 171s 139ms/step - loss: 0.1606 - mean_squared_error: 0.0050 - val_loss: 0.1133 - val_mean_squared_error: 0.0033\n",
      "Epoch 33/50\n",
      "1230/1230 [==============================] - 172s 140ms/step - loss: 0.1574 - mean_squared_error: 0.0049 - val_loss: 0.1331 - val_mean_squared_error: 0.0045\n",
      "Epoch 34/50\n",
      "1230/1230 [==============================] - 171s 139ms/step - loss: 0.1579 - mean_squared_error: 0.0049 - val_loss: 0.1257 - val_mean_squared_error: 0.0040\n",
      "Epoch 35/50\n",
      "1230/1230 [==============================] - 170s 138ms/step - loss: 0.1544 - mean_squared_error: 0.0047 - val_loss: 0.1103 - val_mean_squared_error: 0.0033\n",
      "Epoch 36/50\n",
      "1230/1230 [==============================] - 169s 137ms/step - loss: 0.1519 - mean_squared_error: 0.0045 - val_loss: 0.1199 - val_mean_squared_error: 0.0035\n",
      "Epoch 37/50\n",
      "1230/1230 [==============================] - 168s 136ms/step - loss: 0.1502 - mean_squared_error: 0.0044 - val_loss: 0.1173 - val_mean_squared_error: 0.0036\n",
      "Epoch 38/50\n",
      "1230/1230 [==============================] - 169s 138ms/step - loss: 0.1475 - mean_squared_error: 0.0043 - val_loss: 0.1068 - val_mean_squared_error: 0.0033\n",
      "Epoch 39/50\n",
      "1230/1230 [==============================] - 168s 136ms/step - loss: 0.1463 - mean_squared_error: 0.0042 - val_loss: 0.1228 - val_mean_squared_error: 0.0035\n",
      "Epoch 40/50\n",
      "1230/1230 [==============================] - 168s 136ms/step - loss: 0.1441 - mean_squared_error: 0.0041 - val_loss: 0.1153 - val_mean_squared_error: 0.0039\n",
      "Epoch 41/50\n",
      "1230/1230 [==============================] - 168s 137ms/step - loss: 0.1430 - mean_squared_error: 0.0040 - val_loss: 0.1102 - val_mean_squared_error: 0.0033\n",
      "Epoch 42/50\n",
      "1230/1230 [==============================] - 169s 137ms/step - loss: 0.1411 - mean_squared_error: 0.0039 - val_loss: 0.1218 - val_mean_squared_error: 0.0036\n",
      "Epoch 43/50\n",
      "1230/1230 [==============================] - 173s 140ms/step - loss: 0.1395 - mean_squared_error: 0.0039 - val_loss: 0.1013 - val_mean_squared_error: 0.0029\n",
      "Epoch 44/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.1376 - mean_squared_error: 0.0037 - val_loss: 0.1218 - val_mean_squared_error: 0.0034\n",
      "Epoch 45/50\n",
      "1230/1230 [==============================] - 160s 130ms/step - loss: 0.1364 - mean_squared_error: 0.0037 - val_loss: 0.1066 - val_mean_squared_error: 0.0029\n",
      "Epoch 46/50\n",
      "1230/1230 [==============================] - 160s 130ms/step - loss: 0.1351 - mean_squared_error: 0.0036 - val_loss: 0.0945 - val_mean_squared_error: 0.0027\n",
      "Epoch 47/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.1331 - mean_squared_error: 0.0035 - val_loss: 0.1048 - val_mean_squared_error: 0.0031\n",
      "Epoch 48/50\n",
      "1230/1230 [==============================] - 160s 130ms/step - loss: 0.1321 - mean_squared_error: 0.0035 - val_loss: 0.0961 - val_mean_squared_error: 0.0027\n",
      "Epoch 49/50\n",
      "1230/1230 [==============================] - 160s 130ms/step - loss: 0.1315 - mean_squared_error: 0.0034 - val_loss: 0.0962 - val_mean_squared_error: 0.0027\n",
      "Epoch 50/50\n",
      "1230/1230 [==============================] - 161s 131ms/step - loss: 0.1307 - mean_squared_error: 0.0034 - val_loss: 0.0951 - val_mean_squared_error: 0.0026\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op while saving (showing 5 of 8). These functions will not be directly callable after loading.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: model/model_21_06_03/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: model/model_21_06_03/assets\n"
     ]
    }
   ],
   "source": [
    "model_path = \"model\"\n",
    "\n",
    "# generator for model.fit()\n",
    "def generate(batch_size, s='train'):\n",
    "    batch_index = 0\n",
    "    while True:\n",
    "        if s == 'train':\n",
    "            data = training_data\n",
    "        elif s == 'val':\n",
    "            data = validation_data\n",
    "        else:\n",
    "            print('check operation mode parameter')\n",
    "            return None\n",
    "\n",
    "        batch_x, batch_y = [], []\n",
    "        for i in range(batch_size):\n",
    "            index = batch_index + i\n",
    "            if index >= len(data):\n",
    "                print(\"WARNING: check if batch size is factor of data size\")\n",
    "                index = index % len(data)\n",
    "            batch_x.append((data[index][0].astype(float) - 127.5) / 127.5)\n",
    "            batch_y.append(data[index][1].astype(float).flatten())\n",
    "        \n",
    "        batch_x = np.asarray(batch_x)\n",
    "        batch_y = np.asarray(batch_y)\n",
    "\n",
    "        yield batch_x, batch_y\n",
    "        batch_index += batch_size\n",
    "\n",
    "        # reset batch_index if next epoch\n",
    "        if s=='train' and batch_index >= len(training_data):\n",
    "            batch_index = 0\n",
    "        elif s=='val' and batch_index >= len(validation_data):\n",
    "            batch_index = 0\n",
    "\n",
    "if VALIDATION:\n",
    "    history = model.fit(generate(BATCH_SIZE_TRAIN, s='train'), epochs=EPOCHS, steps_per_epoch = len(training_data)//BATCH_SIZE_TRAIN, validation_data = generate(batch_size_val, s='val'), validation_steps = len(validation_data)//batch_size_val)\n",
    "else:\n",
    "    history = model.fit(generate(BATCH_SIZE_TRAIN, s='train'), epochs=EPOCHS, steps_per_epoch = len(training_data)//BATCH_SIZE_TRAIN)\n",
    "\n",
    "current_time = datetime.now().strftime(\"%Y-%m-%d_%H-%M\")\n",
    "model.save(os.path.join(model_path, \"tf_model_\" + current_time), save_format='tf')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(history.history['loss'])\n",
    "if VALIDATION:\n",
    "    plt.plot(history.history['val_loss'])\n",
    "plt.title('model loss')\n",
    "plt.ylabel('loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'val'], loc='upper left')\n",
    "plt.show()\n",
    "\n",
    "# visualize mse\n",
    "plt.plot(history.history['mean_squared_error'])\n",
    "if VALIDATION:\n",
    "    plt.plot(history.history['val_mean_squared_error'])\n",
    "plt.title('model mse')\n",
    "plt.ylabel('mean_squared_error')\n",
    "plt.xlabel('epoch')\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.10.6 64-bit",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.10 (default, Jun 22 2022, 20:18:18) \n[GCC 9.4.0]"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "916dbcbb3f70747c44a77c7bcd40155683ae19c65e1c03b4aa3499c5328201f1"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}