MobileNetLSTM.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "name": "MobileNetLSTM.ipynb",
      "provenance": [],
      "machine_shape": "hm",
      "authorship_tag": "ABX9TyMwfYSzOvQNm166QmcQOrn9",
      "include_colab_link": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/yoon-303/LipService/blob/main/MobileNetLSTM.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "16GDW75_KqcG"
      },
      "source": [
        "import numpy as np\n",
        "from sklearn.model_selection import train_test_split\n",
        "from keras.callbacks import EarlyStopping\n",
        "import tensorflow as tf\n",
        "from tensorflow import keras\n",
        "from tensorflow.keras import layers\n",
        "from keras.models import Sequential\n",
        "from keras import models, layers, activations\n",
        "from keras.layers import Conv2D, TimeDistributed, BatchNormalization, MaxPooling2D, Flatten, Bidirectional, Dense,Dropout,Flatten"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "E4dUbzT_KrJZ",
        "outputId": "de373730-9a73-4395-f4f1-a724e6f87d46"
      },
      "source": [
        "from google.colab import drive\n",
        "\n",
        "drive.mount('/content/gdrive')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Drive already mounted at /content/gdrive; to attempt to forcibly remount, call drive.mount(\"/content/gdrive\", force_remount=True).\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "5CoSZVaaKs7A",
        "outputId": "d9f3de18-be1b-4f88-a7e5-55d73df97d02"
      },
      "source": [
        "x_data = np.load('/content/gdrive/My Drive/jolp/face/face_x_data.npy') # x_save.npy\n",
        "y_data = np.load('/content/gdrive/My Drive/jolp/face/face_y_data.npy') # y_data.npy\n",
        "\n",
        "\n",
        "\n",
        "y_data = keras.utils.to_categorical(y_data, num_classes=None)\n",
        "#X_train, X_test, Y_train, Y_test = train_test_split(x_data, y_data, test_size = 0.2, shuffle = True, stratify=y_data)\n",
        "print(x_data.shape, y_data.shape)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(3000, 27, 64, 64, 3) (3000, 11)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ftDjQ4RwKzDs",
        "outputId": "bb0b3781-2021-4d55-c38a-f6a7daee48f0"
      },
      "source": [
        "# Train과 Test 분리\n",
        "X_train, X_test, Y_train, Y_test = train_test_split(x_data, y_data, test_size = 0.2, shuffle = True, stratify=y_data)\\\n",
        "#X_train, X_test, Y_train, Y_test = train_test_split(x_data, y_data, test_size = 0.2, shuffle = False)\n",
        "print(\"x_train: {}, X_test: {}\".format(X_train.shape, X_test.shape))\n",
        "print(\"Y_train: {}, Y_test: {}\".format(Y_train.shape, Y_test.shape))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "x_train: (2400, 27, 64, 64, 3), X_test: (600, 27, 64, 64, 3)\n",
            "Y_train: (2400, 11), Y_test: (600, 11)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VyeW-Rh-K3Tn"
      },
      "source": [
        "from tensorflow.keras.applications.mobilenet import MobileNet\n",
        "from tensorflow.keras.layers import Input\n",
        "\n",
        "\n",
        "# this could also be the output a different Keras model or layer\n",
        "#input_tensor = Input(shape=(224, 224, 3))\n",
        "\n",
        "#model = InceptionV3(input_tensor=input_tensor, weights='imagenet', include_top=True)\n",
        "\n",
        "class MobileNet_LSTM(keras.Model):\n",
        "    def __init__(self, num_classes):\n",
        "        super(MobileNet_LSTM, self).__init__()\n",
        "        #self.BZ = input_shape[0]\n",
        "        #self.frame = input_shape[1]\n",
        "\n",
        "        self.mobile = TimeDistributed(MobileNet(weights='imagenet', include_top=False))\n",
        "        #self.vgg.trainable = False\n",
        "\n",
        "        self.bilstm = layers.Bidirectional(layers.LSTM(256), merge_mode='concat')\n",
        "        self.dense = layers.Dense(num_classes, activation='softmax')\n",
        "        self.max = TimeDistributed(MaxPooling2D(pool_size=(2, 2)))\n",
        "        self.dropout = layers.Dropout(0.2)\n",
        "        self.flat= TimeDistributed(Flatten())\n",
        "\n",
        "    def call(self, x):\n",
        "        x = self.mobile(x)\n",
        "        x = self.max(x)\n",
        "        x = self.dropout(x)      \n",
        "        x = self.flat(x)\n",
        "        x = self.dropout(x)\n",
        "        x = self.bilstm(x)\n",
        "        return self.dense(x)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YZzd4UlKK5CE"
      },
      "source": [
        "def get_model():\n",
        "    return MobileNet_LSTM(num_classes =11)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "WcjU8VfsK5EV",
        "outputId": "f663f425-0fab-47b9-e1ce-f611a2cffeab"
      },
      "source": [
        "tf.config.experimental_run_functions_eagerly(True)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From <ipython-input-7-bdb3352f611a>:1: experimental_run_functions_eagerly (from tensorflow.python.eager.def_function) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Use `tf.config.run_functions_eagerly` instead of the experimental version.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "kgInUWt1K8MX",
        "outputId": "d0a71b0b-183f-4172-de9d-b0965423ad7d"
      },
      "source": [
        "\n",
        "model = get_model()\n",
        "model.compile(optimizer = 'adam', loss = 'categorical_crossentropy', metrics = ['accuracy'])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:`input_shape` is undefined or non-square, or `rows` is not in [128, 160, 192, 224]. Weights for input shape (224, 224) will be loaded as the default.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "mxaZmL_FK8Qf",
        "outputId": "588e7ee7-db89-48c3-8cec-c8ec5cc623cc"
      },
      "source": [
        "es = EarlyStopping(monitor = 'val_loss', patience = 10)\n",
        "history = model.fit(X_train, Y_train, batch_size= 32, epochs= 100, verbose = 1, validation_split = 0.2, callbacks=[es])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/tensorflow/python/data/ops/dataset_ops.py:3504: UserWarning: Even though the tf.config.experimental_run_functions_eagerly option is set, this option does not apply to tf.data functions. tf.data functions are still traced and executed as graphs.\n",
            "  \"Even though the tf.config.experimental_run_functions_eagerly \"\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/100\n",
            "60/60 [==============================] - 30s 226ms/step - loss: 2.4494 - accuracy: 0.0891 - val_loss: 2.3419 - val_accuracy: 0.0708\n",
            "Epoch 2/100\n",
            "60/60 [==============================] - 12s 207ms/step - loss: 2.2921 - accuracy: 0.1275 - val_loss: 2.3185 - val_accuracy: 0.1187\n",
            "Epoch 3/100\n",
            "60/60 [==============================] - 12s 208ms/step - loss: 2.2607 - accuracy: 0.1619 - val_loss: 2.3075 - val_accuracy: 0.1250\n",
            "Epoch 4/100\n",
            "60/60 [==============================] - 13s 209ms/step - loss: 2.1919 - accuracy: 0.1794 - val_loss: 2.3137 - val_accuracy: 0.1229\n",
            "Epoch 5/100\n",
            "60/60 [==============================] - 12s 206ms/step - loss: 2.1182 - accuracy: 0.2266 - val_loss: 2.2258 - val_accuracy: 0.1521\n",
            "Epoch 6/100\n",
            "60/60 [==============================] - 12s 209ms/step - loss: 2.0219 - accuracy: 0.2721 - val_loss: 2.4668 - val_accuracy: 0.1750\n",
            "Epoch 7/100\n",
            "60/60 [==============================] - 12s 206ms/step - loss: 1.9529 - accuracy: 0.2775 - val_loss: 2.5796 - val_accuracy: 0.1000\n",
            "Epoch 8/100\n",
            "60/60 [==============================] - 12s 208ms/step - loss: 1.8590 - accuracy: 0.3483 - val_loss: 2.3070 - val_accuracy: 0.1958\n",
            "Epoch 9/100\n",
            "60/60 [==============================] - 13s 209ms/step - loss: 1.6871 - accuracy: 0.4094 - val_loss: 3.0564 - val_accuracy: 0.1333\n",
            "Epoch 10/100\n",
            "60/60 [==============================] - 12s 206ms/step - loss: 1.4888 - accuracy: 0.5053 - val_loss: 2.6356 - val_accuracy: 0.1792\n",
            "Epoch 11/100\n",
            "60/60 [==============================] - 12s 208ms/step - loss: 1.4308 - accuracy: 0.5133 - val_loss: 2.2687 - val_accuracy: 0.2333\n",
            "Epoch 12/100\n",
            "60/60 [==============================] - 12s 207ms/step - loss: 1.1723 - accuracy: 0.6184 - val_loss: 2.7549 - val_accuracy: 0.2000\n",
            "Epoch 13/100\n",
            "60/60 [==============================] - 12s 208ms/step - loss: 1.0285 - accuracy: 0.6788 - val_loss: 2.2522 - val_accuracy: 0.2958\n",
            "Epoch 14/100\n",
            "60/60 [==============================] - 12s 206ms/step - loss: 0.9533 - accuracy: 0.7101 - val_loss: 2.8780 - val_accuracy: 0.2167\n",
            "Epoch 15/100\n",
            "60/60 [==============================] - 12s 209ms/step - loss: 0.8748 - accuracy: 0.7334 - val_loss: 2.2894 - val_accuracy: 0.3042\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "z-hY8020LDrS",
        "outputId": "0397b7c3-fc4d-4304-98dc-7e100ff3479a"
      },
      "source": [
        "history.history"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{'accuracy': [0.09375,\n",
              "  0.12083332985639572,\n",
              "  0.15052083134651184,\n",
              "  0.1770833283662796,\n",
              "  0.2083333283662796,\n",
              "  0.2645833194255829,\n",
              "  0.2874999940395355,\n",
              "  0.3505208194255829,\n",
              "  0.4026041626930237,\n",
              "  0.4661458432674408,\n",
              "  0.5182291865348816,\n",
              "  0.573437511920929,\n",
              "  0.640625,\n",
              "  0.6875,\n",
              "  0.7020833492279053],\n",
              " 'loss': [2.390050172805786,\n",
              "  2.3054513931274414,\n",
              "  2.2700226306915283,\n",
              "  2.2069694995880127,\n",
              "  2.152264356613159,\n",
              "  2.0157082080841064,\n",
              "  1.9452288150787354,\n",
              "  1.8532741069793701,\n",
              "  1.7030147314071655,\n",
              "  1.5469528436660767,\n",
              "  1.4371014833450317,\n",
              "  1.2581655979156494,\n",
              "  1.1025843620300293,\n",
              "  0.9785966277122498,\n",
              "  0.92221999168396],\n",
              " 'val_accuracy': [0.07083333283662796,\n",
              "  0.11874999850988388,\n",
              "  0.125,\n",
              "  0.12291666865348816,\n",
              "  0.15208333730697632,\n",
              "  0.17499999701976776,\n",
              "  0.10000000149011612,\n",
              "  0.19583334028720856,\n",
              "  0.13333334028720856,\n",
              "  0.17916665971279144,\n",
              "  0.23333333432674408,\n",
              "  0.20000000298023224,\n",
              "  0.2958333194255829,\n",
              "  0.21666666865348816,\n",
              "  0.30416667461395264],\n",
              " 'val_loss': [2.3419227600097656,\n",
              "  2.3184635639190674,\n",
              "  2.307522773742676,\n",
              "  2.313732147216797,\n",
              "  2.2258269786834717,\n",
              "  2.466785430908203,\n",
              "  2.5796444416046143,\n",
              "  2.3069965839385986,\n",
              "  3.056405544281006,\n",
              "  2.6355557441711426,\n",
              "  2.268681764602661,\n",
              "  2.754866123199463,\n",
              "  2.2522494792938232,\n",
              "  2.878035545349121,\n",
              "  2.289443254470825]}"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 12
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 590
        },
        "id": "rIVkh1U-LGd3",
        "outputId": "5ab5f775-1e65-48e4-bb8f-e5ba8f7aca6b"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "# list all data in history\n",
        "print(history.history.keys())\n",
        "# summarize history for accuracy\n",
        "plt.plot(history.history['accuracy'])\n",
        "plt.plot(history.history['val_accuracy'])\n",
        "plt.title('model accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "# summarize history for loss\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "plt.title('model loss')\n",
        "plt.ylabel('loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "dict_keys(['loss', 'accuracy', 'val_loss', 'val_accuracy'])\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KAGquR3OLmm3",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "de796c71-ab3d-41c8-f1ab-486af246969d"
      },
      "source": [
        "score = model.evaluate(X_test, Y_test, batch_size = 32, verbose = 1)\n",
        "print(\"acc:\", score[1], \"loss:\", score[0])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\r 1/19 [>.............................] - ETA: 2s - loss: 2.1825 - accuracy: 0.2500"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.7/dist-packages/tensorflow/python/data/ops/dataset_ops.py:3504: UserWarning: Even though the tf.config.experimental_run_functions_eagerly option is set, this option does not apply to tf.data functions. tf.data functions are still traced and executed as graphs.\n",
            "  \"Even though the tf.config.experimental_run_functions_eagerly \"\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "19/19 [==============================] - 2s 89ms/step - loss: 2.3615 - accuracy: 0.2883\n",
            "acc: 0.28833332657814026 loss: 2.3615081310272217\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "66GoTxGW5Gtm"
      },
      "source": [
        "model.save_weights('mobilenet_lstm.h5')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dfGVVeJ15UCN"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}