imitation_classify_CAE_v2.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "# Pandas and Numpy for data structures and util fucntions\n",
    "import scipy as sp\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from numpy.random import rand\n",
    "pd.options.display.max_colwidth = 600\n",
    "\n",
    "# Scikit Imports\n",
    "from sklearn import preprocessing\n",
    "from sklearn.metrics import roc_curve, auc, precision_recall_curve\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "import cnn_utils as utils\n",
    "from model_evaluation_utils import get_metrics\n",
    "\n",
    "# Matplot Imports\n",
    "import matplotlib.pyplot as plt\n",
    "params = {'legend.fontsize': 'x-large',\n",
    "          'figure.figsize': (15, 5),\n",
    "          'axes.labelsize': 'x-large',\n",
    "          'axes.titlesize':'x-large',\n",
    "          'xtick.labelsize':'x-large',\n",
    "          'ytick.labelsize':'x-large'}\n",
    "\n",
    "plt.rcParams.update(params)\n",
    "%matplotlib inline\n",
    "\n",
    "# pandas display data frames as tables\n",
    "from IPython.display import display, HTML\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "np.random.seed(1337)\n",
    "from keras import callbacks\n",
    "from keras import optimizers\n",
    "#from keras.datasets import cifar10\n",
    "from keras.engine import Model\n",
    "from keras.applications import vgg16 as vgg\n",
    "from keras.layers import Dropout, Flatten, Dense, GlobalAveragePooling2D,BatchNormalization\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "from keras.utils import np_utils\n",
    "from keras.models import Sequential,  Model, model_from_json\n",
    "from keras.optimizers import RMSprop,Adam,SGD\n",
    "from keras.layers import Activation, Dropout, Flatten, Dense, GlobalMaxPooling2D, Conv2D, MaxPooling2D\n",
    "from keras.callbacks import CSVLogger\n",
    "from keras.wrappers.scikit_learn import KerasClassifier\n",
    "from livelossplot.keras import PlotLossesCallback\n",
    "import efficientnet.keras as efn\n",
    "import h5py, glob, re\n",
    "import os"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of training samples:  1841\n",
      "Number of test samples:  527\n",
      "y = 11\n",
      "number of training examples = 1841\n",
      "number of test examples = 527\n",
      "X_train shape: (1841, 128, 128, 3)\n",
      "Y_train shape: (1841, 20)\n",
      "X_test shape: (527, 128, 128, 3)\n",
      "Y_test shape: (527, 20)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import h5py\n",
    "hdf5_path = 'native vs imitation/All_imitation.hdf5' \n",
    "#hdf5_path = 'tobb_dataset_128x128.hdf5'\n",
    "dataset = h5py.File(hdf5_path, \"r\")\n",
    "\n",
    "\n",
    "\n",
    "Y_test_orig = np.array(dataset[\"test_labels\"])\n",
    "Y_train_orig = np.array(dataset[\"train_labels\"])\n",
    "Y_val_orig = np.array(dataset[\"valid_labels\"])\n",
    "\n",
    "X_train_orig = np.array(dataset[\"train_img\"])\n",
    "X_test_orig = np.array(dataset[\"test_img\"])\n",
    "X_val_orig = np.array(dataset[\"valid_img\"])\n",
    "\n",
    "print('Number of training samples: ', len(X_train_orig))\n",
    "print('Number of test samples: ', len(X_test_orig))\n",
    "\n",
    "# Example of a picture\n",
    "index = 14\n",
    "plt.imshow(X_train_orig[index])\n",
    "print (\"y = \" + str(np.squeeze(Y_train_orig[index])))\n",
    "\n",
    "\n",
    "X_train = X_train_orig/255.\n",
    "X_test = X_test_orig/255.\n",
    "X_val  = X_val_orig/255\n",
    "#X_train, X_val, y_train, y_val = train_test_split(X_train, \n",
    "#                                                  Y_train_orig, \n",
    "#                                                  test_size=0.15, \n",
    "#                                                 stratify=np.array(Y_train_orig), \n",
    "#                                                  random_state=42)\n",
    "\n",
    "\n",
    "NUM_CLASSES=20;\n",
    "\n",
    "Y_train = np_utils.to_categorical(Y_train_orig, NUM_CLASSES)\n",
    "Y_val = np_utils.to_categorical(Y_val_orig, NUM_CLASSES)\n",
    "Y_test = np_utils.to_categorical(Y_test_orig, NUM_CLASSES)\n",
    "print (\"number of training examples = \" + str(X_train.shape[0]))\n",
    "print (\"number of test examples = \" + str(X_test.shape[0]))\n",
    "print (\"X_train shape: \" + str(X_train.shape))\n",
    "print (\"Y_train shape: \" + str(Y_train.shape))\n",
    "print (\"X_test shape: \" + str(X_test.shape))\n",
    "print (\"Y_test shape: \" + str(Y_test.shape))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<bound method Network.summary of <keras.engine.sequential.Sequential object at 0x7f4cd435eba8>>"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "radar = 0 # '0' 4ghz, '1' 10ghz, '2' 24 ghz, or '3' 77\n",
    "# model_addr = glob.glob('models/88x120 CAE/New Models Excel/*.json')\n",
    "# weight_addr = glob.glob('models/88x120 CAE/New Models Excel/*.h5')\n",
    "# model_addr = glob.glob('models/GAN/*.json')\n",
    "# weight_addr = glob.glob('models/GAN/*.h5')\n",
    "model_addr = glob.glob('native vs imitation/*.json')\n",
    "#weight_addr = glob.glob('models/128x128/Auto/Selected Models/*.h5')\n",
    "#print(model_addr)\n",
    "#print(weight_addr)\n",
    "model_file = model_addr[radar]\n",
    "#weights_file = weight_addr[radar]\n",
    "\n",
    "json_file = open(model_file, 'r')\n",
    "loaded_model_json = json_file.read()\n",
    "json_file.close()\n",
    "loaded_model = model_from_json(loaded_model_json)\n",
    "opt = SGD(lr=0.0001, momentum=0.9, decay=1e-6)\n",
    "loaded_model.compile(optimizer=opt, loss='categorical_crossentropy', metrics=['accuracy'])\n",
    "\n",
    "#history = loaded_model.fit(X_train, Y_train, epochs = 50, validation_data = (X_test,Y_test),  batch_size = 16, verbose=1)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_79\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "model_80 (Model)             (None, 4, 4, 128)         2967040   \n",
      "_________________________________________________________________\n",
      "flatten_79 (Flatten)         (None, 2048)              0         \n",
      "_________________________________________________________________\n",
      "dense_235 (Dense)            (None, 256)               524544    \n",
      "_________________________________________________________________\n",
      "dropout_157 (Dropout)        (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dense_236 (Dense)            (None, 128)               32896     \n",
      "_________________________________________________________________\n",
      "dropout_158 (Dropout)        (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "dense_237 (Dense)            (None, 11)                1419      \n",
      "_________________________________________________________________\n",
      "activation_79 (Activation)   (None, 11)                0         \n",
      "=================================================================\n",
      "Total params: 3,525,899\n",
      "Trainable params: 3,525,899\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "loaded_model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_3\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "model_80 (Model)             (None, 4, 4, 128)         2967040   \n",
      "_________________________________________________________________\n",
      "flatten_79 (Flatten)         (None, 2048)              0         \n",
      "_________________________________________________________________\n",
      "dense_235 (Dense)            (None, 256)               524544    \n",
      "_________________________________________________________________\n",
      "dropout_157 (Dropout)        (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dense_236 (Dense)            (None, 128)               32896     \n",
      "_________________________________________________________________\n",
      "dropout_158 (Dropout)        (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 20)                2580      \n",
      "_________________________________________________________________\n",
      "activation_2 (Activation)    (None, 20)                0         \n",
      "=================================================================\n",
      "Total params: 3,527,060\n",
      "Trainable params: 3,527,060\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Train on 1841 samples, validate on 527 samples\n",
      "Epoch 1/50\n",
      "1841/1841 [==============================] - 261s 142ms/step - loss: 3.0115 - accuracy: 0.0489 - val_loss: 2.9791 - val_accuracy: 0.0873\n",
      "Epoch 2/50\n",
      "1841/1841 [==============================] - 262s 142ms/step - loss: 2.9891 - accuracy: 0.0663 - val_loss: 2.9717 - val_accuracy: 0.1082\n",
      "Epoch 3/50\n",
      "1841/1841 [==============================] - 260s 141ms/step - loss: 2.9851 - accuracy: 0.0625 - val_loss: 2.9680 - val_accuracy: 0.1025\n",
      "Epoch 4/50\n",
      "1841/1841 [==============================] - 259s 141ms/step - loss: 2.9788 - accuracy: 0.0771 - val_loss: 2.9623 - val_accuracy: 0.1063\n",
      "Epoch 5/50\n",
      "1841/1841 [==============================] - 261s 142ms/step - loss: 2.9727 - accuracy: 0.0809 - val_loss: 2.9603 - val_accuracy: 0.1176\n",
      "Epoch 6/50\n",
      "1841/1841 [==============================] - 262s 142ms/step - loss: 2.9804 - accuracy: 0.0777 - val_loss: 2.9561 - val_accuracy: 0.1271\n",
      "Epoch 7/50\n",
      "1841/1841 [==============================] - 263s 143ms/step - loss: 2.9721 - accuracy: 0.0788 - val_loss: 2.9520 - val_accuracy: 0.1290\n",
      "Epoch 8/50\n",
      "1841/1841 [==============================] - 262s 142ms/step - loss: 2.9680 - accuracy: 0.0880 - val_loss: 2.9489 - val_accuracy: 0.1233\n",
      "Epoch 9/50\n",
      "1841/1841 [==============================] - 263s 143ms/step - loss: 2.9681 - accuracy: 0.0896 - val_loss: 2.9543 - val_accuracy: 0.1176\n",
      "Epoch 10/50\n",
      "1841/1841 [==============================] - 262s 142ms/step - loss: 2.9738 - accuracy: 0.0771 - val_loss: 2.9446 - val_accuracy: 0.1385\n",
      "Epoch 11/50\n",
      "1841/1841 [==============================] - 225s 122ms/step - loss: 2.9645 - accuracy: 0.0864 - val_loss: 2.9407 - val_accuracy: 0.1233\n",
      "Epoch 12/50\n",
      "1841/1841 [==============================] - 224s 122ms/step - loss: 2.9706 - accuracy: 0.0777 - val_loss: 2.9378 - val_accuracy: 0.1195\n",
      "Epoch 13/50\n",
      " 720/1841 [==========>...................] - ETA: 2:12 - loss: 2.9604 - accuracy: 0.1000"
     ]
    }
   ],
   "source": [
    "\n",
    "\n",
    "model = Sequential()\n",
    "# for layer in loaded_model.layers[:-2]: # go through until last layer\n",
    "#     model.add(layer)\n",
    "# model.add(Dense(20, activation='relu'))\n",
    "# model.add(Activation('softmax'))\n",
    "im_width = 128\n",
    "im_height = 128\n",
    "inChannel = 3\n",
    "input_img = Input(shape = (im_width, im_height, inChannel)) \n",
    "model2 = Model(input_img , outputs=loaded_model.layers[:-2].output)\n",
    "model.add(model2)\n",
    "model.add(Dense(NUM_CLASSES))\n",
    "model.add(Activation('softmax'))\n",
    "optim = Adam(lr=0.0002)\n",
    "model.summary()\n",
    "# model.compile(optimizer=opt, loss='categorical_crossentropy', metrics=['accuracy'])\n",
    "model.compile(optimizer=optim, loss='categorical_crossentropy', metrics=['accuracy'])\n",
    "\n",
    "\n",
    "history = model.fit(X_train, Y_train, epochs = 50, validation_data = (X_test,Y_test),  batch_size = 16, verbose=1)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_2\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "model_80 (Model)             (None, 4, 4, 128)         2967040   \n",
      "_________________________________________________________________\n",
      "flatten_79 (Flatten)         (None, 2048)              0         \n",
      "_________________________________________________________________\n",
      "dense_235 (Dense)            (None, 256)               524544    \n",
      "_________________________________________________________________\n",
      "dropout_157 (Dropout)        (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dense_236 (Dense)            (None, 128)               32896     \n",
      "_________________________________________________________________\n",
      "dropout_158 (Dropout)        (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 20)                2580      \n",
      "_________________________________________________________________\n",
      "activation_1 (Activation)    (None, 20)                0         \n",
      "=================================================================\n",
      "Total params: 3,527,060\n",
      "Trainable params: 3,527,060\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[10  0  0  0  2  0  0  1  0  0  0  0  0  5  0  0  0  0  0  0]\n",
      " [ 0  0  0  0  8  0  7  3  0  0  0  1  0 10  0  0  0  0  0  0]\n",
      " [ 0  0  0  0  0  0  3  1  0  0  0  0  0 15  0  0  0  0  0  0]\n",
      " [ 0  0  0  0  2  0  9  1  0  0  0  1  0  5  0  0  0  0  7  0]\n",
      " [ 0  0  0  0 17  0  6  0  0  0  0  1  0 13  0  0  0  0  0  0]\n",
      " [ 0  0  0  0  3  0 12  0  0  0  0  0  0  6  0  0  0  0  3  0]\n",
      " [ 0  0  0  0  0  0 10  1  0  0  0  0  0  3  0  0  0  0  6  0]\n",
      " [ 0  0  0  0  1  0  5  1  0  0  0  0  0 24  0  0  0  0  0  0]\n",
      " [ 1  0  0  0  2  0  0  2  0  0  0  0  0 24  0  0  0  0  0  0]\n",
      " [ 0  0  0  0  1  0 14  0  0  0  0  2  0  3  0  0  0  0  3  0]\n",
      " [ 0  0  0  0  2  0 14  2  0  0  0  2  0 18  0  0  0  0  0  0]\n",
      " [ 0  0  0  0  1  0  4  6  0  0  0  2  0 16  0  0  0  0  0  0]\n",
      " [ 0  0  0  0  0  0 14  3  0  0  0  3  0  7  0  0  0  0  7  0]\n",
      " [ 1  0  0  0  5  0  5  0  0  0  0  2  0 13  0  0  0  0  1  0]\n",
      " [ 0  0  0  0  8  0  8  0  0  0  0  0  0 12  0  0  0  0  0  0]\n",
      " [ 4  0  0  0  0  0  1  3  0  0  0  0  0 14  0  0  0  0  0  0]\n",
      " [ 1  0  0  0  3  0 15  4  0  0  0  1  0  2  0  0  0  0  0  0]\n",
      " [ 0  0  0  0  0  0 10  4  0  0  0  1  0  8  0  0  0  0  2  0]\n",
      " [ 0  0  0  0  0  0 13  0  0  0  0  0  0  3  0  0  0  0  4  0]\n",
      " [ 0  0  0  0  3  0  3  1  0  0  0  0  0  9  0  1  0  0  6  0]]\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "\n",
    "tt= model.predict(X_test)\n",
    "cm=confusion_matrix(np.argmax(Y_test,axis=1), np.argmax(tt,axis=1))\n",
    "print(cm)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}