cnn-OD.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "ename": "ModuleNotFoundError",
     "evalue": "No module named 'keras'",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mModuleNotFoundError\u001b[0m                       Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-1-3ca079507c8c>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0;32mfrom\u001b[0m \u001b[0mkeras\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdatasets\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mcifar10\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      2\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0mkeras\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmodels\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mSequential\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0mkeras\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mlayers\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mDense\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0mkeras\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mlayers\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mDropout\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0mkeras\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mlayers\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mFlatten\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mModuleNotFoundError\u001b[0m: No module named 'keras'"
     ]
    }
   ],
   "source": [
    "from keras.datasets import cifar10\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "from keras.layers import Dropout\n",
    "from keras.layers import Flatten\n",
    "from keras.constraints import maxnorm\n",
    "from keras.optimizers import SGD\n",
    "from keras.layers.convolutional import Conv2D\n",
    "from keras.layers.convolutional import MaxPooling2D\n",
    "from keras.utils import np_utils\n",
    "from matplotlib import pyplot"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# load data\n",
    "(X_train, y_train), (X_test, y_test) = cifar10.load_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 9 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# create a grid of 3x3 images\n",
    "for i in range(0, 9):\n",
    "\tpyplot.subplot(330 + 1 + i)\n",
    "\tpyplot.imshow(X_train[i])\n",
    "# show the plot\n",
    "pyplot.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# normalize inputs from 0-255 to 0.0-1.0\n",
    "X_train = X_train.astype('float32')\n",
    "X_test = X_test.astype('float32')\n",
    "X_train = X_train / 255.0\n",
    "X_test = X_test / 255.0\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# one hot encode outputs\n",
    "y_train = np_utils.to_categorical(y_train)\n",
    "y_test = np_utils.to_categorical(y_test)\n",
    "num_classes = y_test.shape[1]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING: Logging before flag parsing goes to stderr.\n",
      "W1110 20:36:44.318284 140381610207040 deprecation_wrapper.py:119] From /home/alim/anaconda3/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:74: The name tf.get_default_graph is deprecated. Please use tf.compat.v1.get_default_graph instead.\n",
      "\n",
      "W1110 20:36:44.343456 140381610207040 deprecation_wrapper.py:119] From /home/alim/anaconda3/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:517: The name tf.placeholder is deprecated. Please use tf.compat.v1.placeholder instead.\n",
      "\n",
      "W1110 20:36:44.348010 140381610207040 deprecation_wrapper.py:119] From /home/alim/anaconda3/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:4138: The name tf.random_uniform is deprecated. Please use tf.random.uniform instead.\n",
      "\n",
      "W1110 20:36:44.370718 140381610207040 deprecation_wrapper.py:119] From /home/alim/anaconda3/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:133: The name tf.placeholder_with_default is deprecated. Please use tf.compat.v1.placeholder_with_default instead.\n",
      "\n",
      "W1110 20:36:44.381223 140381610207040 deprecation.py:506] From /home/alim/anaconda3/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:3445: calling dropout (from tensorflow.python.ops.nn_ops) with keep_prob is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Please use `rate` instead of `keep_prob`. Rate should be set to `rate = 1 - keep_prob`.\n",
      "W1110 20:36:44.422686 140381610207040 deprecation_wrapper.py:119] From /home/alim/anaconda3/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:3976: The name tf.nn.max_pool is deprecated. Please use tf.nn.max_pool2d instead.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "\n",
    "# Create the model\n",
    "model = Sequential()\n",
    "model.add(Conv2D(32, (3, 3), input_shape=(32, 32, 3), padding='same', activation='relu', kernel_constraint=maxnorm(3)))\n",
    "model.add(Dropout(0.2))\n",
    "model.add(Conv2D(32, (3, 3), activation='relu', padding='same', kernel_constraint=maxnorm(3)))\n",
    "model.add(MaxPooling2D())\n",
    "model.add(Flatten())\n",
    "model.add(Dense(512, activation='relu', kernel_constraint=maxnorm(3)))\n",
    "model.add(Dropout(0.5))\n",
    "model.add(Dense(num_classes, activation='softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "W1110 20:37:02.927600 140381610207040 deprecation_wrapper.py:119] From /home/alim/anaconda3/lib/python3.7/site-packages/keras/optimizers.py:790: The name tf.train.Optimizer is deprecated. Please use tf.compat.v1.train.Optimizer instead.\n",
      "\n",
      "W1110 20:37:02.943372 140381610207040 deprecation_wrapper.py:119] From /home/alim/anaconda3/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:3295: The name tf.log is deprecated. Please use tf.math.log instead.\n",
      "\n",
      "W1110 20:37:03.073609 140381610207040 deprecation.py:323] From /home/alim/anaconda3/lib/python3.7/site-packages/tensorflow/python/ops/math_grad.py:1250: add_dispatch_support.<locals>.wrapper (from tensorflow.python.ops.array_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use tf.where in 2.0, which has the same broadcast rule as np.where\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_1 (Conv2D)            (None, 32, 32, 32)        896       \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 32, 32, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 32, 32, 32)        9248      \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 16, 16, 32)        0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 8192)              0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 512)               4194816   \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 10)                5130      \n",
      "=================================================================\n",
      "Total params: 4,210,090\n",
      "Trainable params: 4,210,090\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Train on 50000 samples, validate on 10000 samples\n",
      "Epoch 1/50\n",
      "50000/50000 [==============================] - 273s 5ms/step - loss: 1.7072 - acc: 0.3828 - val_loss: 1.4238 - val_acc: 0.4977\n",
      "Epoch 2/50\n",
      "50000/50000 [==============================] - 272s 5ms/step - loss: 1.3370 - acc: 0.5221 - val_loss: 1.1845 - val_acc: 0.5670\n",
      "Epoch 3/50\n",
      "50000/50000 [==============================] - 277s 6ms/step - loss: 1.1606 - acc: 0.5854 - val_loss: 1.0726 - val_acc: 0.6163\n",
      "Epoch 4/50\n",
      "50000/50000 [==============================] - 276s 6ms/step - loss: 1.0343 - acc: 0.6329 - val_loss: 1.0094 - val_acc: 0.6427\n",
      "Epoch 5/50\n",
      "50000/50000 [==============================] - 275s 5ms/step - loss: 0.9280 - acc: 0.6713 - val_loss: 0.9712 - val_acc: 0.6544\n",
      "Epoch 6/50\n",
      "50000/50000 [==============================] - 277s 6ms/step - loss: 0.8326 - acc: 0.7040 - val_loss: 0.9626 - val_acc: 0.6597\n",
      "Epoch 7/50\n",
      "50000/50000 [==============================] - 278s 6ms/step - loss: 0.7516 - acc: 0.7349 - val_loss: 0.9319 - val_acc: 0.6765\n",
      "Epoch 8/50\n",
      "50000/50000 [==============================] - 276s 6ms/step - loss: 0.6711 - acc: 0.7605 - val_loss: 0.9067 - val_acc: 0.6920\n",
      "Epoch 9/50\n",
      "50000/50000 [==============================] - 274s 5ms/step - loss: 0.6042 - acc: 0.7861 - val_loss: 0.9362 - val_acc: 0.6904\n",
      "Epoch 10/50\n",
      "50000/50000 [==============================] - 272s 5ms/step - loss: 0.5450 - acc: 0.8069 - val_loss: 0.9110 - val_acc: 0.6978\n",
      "Epoch 11/50\n",
      "50000/50000 [==============================] - 267s 5ms/step - loss: 0.4872 - acc: 0.8254 - val_loss: 0.9210 - val_acc: 0.7022\n",
      "Epoch 12/50\n",
      "50000/50000 [==============================] - 257s 5ms/step - loss: 0.4425 - acc: 0.8425 - val_loss: 0.9379 - val_acc: 0.6999\n",
      "Epoch 13/50\n",
      "50000/50000 [==============================] - 258s 5ms/step - loss: 0.3985 - acc: 0.8573 - val_loss: 0.9595 - val_acc: 0.7073\n",
      "Epoch 14/50\n",
      "50000/50000 [==============================] - 276s 6ms/step - loss: 0.3630 - acc: 0.8720 - val_loss: 1.0056 - val_acc: 0.7037\n",
      "Epoch 15/50\n",
      "50000/50000 [==============================] - 274s 5ms/step - loss: 0.3272 - acc: 0.8842 - val_loss: 0.9944 - val_acc: 0.7094\n",
      "Epoch 16/50\n",
      "50000/50000 [==============================] - 275s 5ms/step - loss: 0.3005 - acc: 0.8944 - val_loss: 1.0056 - val_acc: 0.7123\n",
      "Epoch 17/50\n",
      "50000/50000 [==============================] - 277s 6ms/step - loss: 0.2785 - acc: 0.9026 - val_loss: 1.0287 - val_acc: 0.7084\n",
      "Epoch 18/50\n",
      "50000/50000 [==============================] - 276s 6ms/step - loss: 0.2550 - acc: 0.9098 - val_loss: 1.0448 - val_acc: 0.7045\n",
      "Epoch 19/50\n",
      "50000/50000 [==============================] - 271s 5ms/step - loss: 0.2347 - acc: 0.9185 - val_loss: 1.0762 - val_acc: 0.7059\n",
      "Epoch 20/50\n",
      "50000/50000 [==============================] - 270s 5ms/step - loss: 0.2186 - acc: 0.9235 - val_loss: 1.0834 - val_acc: 0.7072\n",
      "Epoch 21/50\n",
      "50000/50000 [==============================] - 271s 5ms/step - loss: 0.2056 - acc: 0.9273 - val_loss: 1.1518 - val_acc: 0.7041\n",
      "Epoch 22/50\n",
      "50000/50000 [==============================] - 271s 5ms/step - loss: 0.1884 - acc: 0.9346 - val_loss: 1.1381 - val_acc: 0.7102\n",
      "Epoch 23/50\n",
      "50000/50000 [==============================] - 258s 5ms/step - loss: 0.1735 - acc: 0.9399 - val_loss: 1.1520 - val_acc: 0.7029\n",
      "Epoch 24/50\n",
      "50000/50000 [==============================] - 251s 5ms/step - loss: 0.1653 - acc: 0.9432 - val_loss: 1.1459 - val_acc: 0.7088\n",
      "Epoch 25/50\n",
      "50000/50000 [==============================] - 250s 5ms/step - loss: 0.1574 - acc: 0.9455 - val_loss: 1.1937 - val_acc: 0.7076\n",
      "Epoch 26/50\n",
      "50000/50000 [==============================] - 251s 5ms/step - loss: 0.1481 - acc: 0.9504 - val_loss: 1.1844 - val_acc: 0.7068\n",
      "Epoch 27/50\n",
      "50000/50000 [==============================] - 251s 5ms/step - loss: 0.1461 - acc: 0.9505 - val_loss: 1.1710 - val_acc: 0.7101\n",
      "Epoch 28/50\n",
      "50000/50000 [==============================] - 251s 5ms/step - loss: 0.1342 - acc: 0.9535 - val_loss: 1.2596 - val_acc: 0.7094\n",
      "Epoch 29/50\n",
      "50000/50000 [==============================] - 262s 5ms/step - loss: 0.1283 - acc: 0.9561 - val_loss: 1.2314 - val_acc: 0.7081\n",
      "Epoch 30/50\n",
      "50000/50000 [==============================] - 272s 5ms/step - loss: 0.1218 - acc: 0.9592 - val_loss: 1.2401 - val_acc: 0.7131\n",
      "Epoch 31/50\n",
      "50000/50000 [==============================] - 270s 5ms/step - loss: 0.1158 - acc: 0.9602 - val_loss: 1.2436 - val_acc: 0.7119\n",
      "Epoch 32/50\n",
      "50000/50000 [==============================] - 271s 5ms/step - loss: 0.1127 - acc: 0.9625 - val_loss: 1.2350 - val_acc: 0.7115\n",
      "Epoch 33/50\n",
      "50000/50000 [==============================] - 274s 5ms/step - loss: 0.1062 - acc: 0.9630 - val_loss: 1.2553 - val_acc: 0.7120\n",
      "Epoch 34/50\n",
      "50000/50000 [==============================] - 280s 6ms/step - loss: 0.1026 - acc: 0.9650 - val_loss: 1.2742 - val_acc: 0.7118\n",
      "Epoch 35/50\n",
      "50000/50000 [==============================] - 309s 6ms/step - loss: 0.1006 - acc: 0.9660 - val_loss: 1.2863 - val_acc: 0.7142\n",
      "Epoch 36/50\n",
      "50000/50000 [==============================] - 302s 6ms/step - loss: 0.0962 - acc: 0.9671 - val_loss: 1.3233 - val_acc: 0.7137\n",
      "Epoch 37/50\n",
      "50000/50000 [==============================] - 269s 5ms/step - loss: 0.0900 - acc: 0.9699 - val_loss: 1.3071 - val_acc: 0.7125\n",
      "Epoch 38/50\n",
      "50000/50000 [==============================] - 340s 7ms/step - loss: 0.0878 - acc: 0.9699 - val_loss: 1.3135 - val_acc: 0.7144\n",
      "Epoch 39/50\n",
      "50000/50000 [==============================] - 327s 7ms/step - loss: 0.0836 - acc: 0.9724 - val_loss: 1.3383 - val_acc: 0.7115\n",
      "Epoch 40/50\n",
      "50000/50000 [==============================] - 272s 5ms/step - loss: 0.0817 - acc: 0.9728 - val_loss: 1.3768 - val_acc: 0.7111\n",
      "Epoch 41/50\n",
      "50000/50000 [==============================] - 274s 5ms/step - loss: 0.0831 - acc: 0.9727 - val_loss: 1.3533 - val_acc: 0.7130\n",
      "Epoch 42/50\n",
      "50000/50000 [==============================] - 273s 5ms/step - loss: 0.0783 - acc: 0.9739 - val_loss: 1.3557 - val_acc: 0.7130\n",
      "Epoch 43/50\n",
      "50000/50000 [==============================] - 272s 5ms/step - loss: 0.0766 - acc: 0.9743 - val_loss: 1.3722 - val_acc: 0.7116\n",
      "Epoch 44/50\n",
      "50000/50000 [==============================] - 273s 5ms/step - loss: 0.0741 - acc: 0.9758 - val_loss: 1.3597 - val_acc: 0.7154\n",
      "Epoch 45/50\n",
      "50000/50000 [==============================] - 273s 5ms/step - loss: 0.0719 - acc: 0.9764 - val_loss: 1.3819 - val_acc: 0.7140\n",
      "Epoch 46/50\n",
      "50000/50000 [==============================] - 272s 5ms/step - loss: 0.0692 - acc: 0.9775 - val_loss: 1.3939 - val_acc: 0.7137\n",
      "Epoch 47/50\n",
      "50000/50000 [==============================] - 272s 5ms/step - loss: 0.0686 - acc: 0.9776 - val_loss: 1.3764 - val_acc: 0.7153\n",
      "Epoch 48/50\n",
      "50000/50000 [==============================] - 274s 5ms/step - loss: 0.0679 - acc: 0.9780 - val_loss: 1.4053 - val_acc: 0.7156\n",
      "Epoch 49/50\n",
      "50000/50000 [==============================] - 272s 5ms/step - loss: 0.0679 - acc: 0.9777 - val_loss: 1.4029 - val_acc: 0.7138\n",
      "Epoch 50/50\n",
      "50000/50000 [==============================] - 272s 5ms/step - loss: 0.0642 - acc: 0.9793 - val_loss: 1.4113 - val_acc: 0.7138\n",
      "Accuracy: 71.38%\n"
     ]
    }
   ],
   "source": [
    "# Compile model\n",
    "epochs = 50\n",
    "lrate = 0.01\n",
    "decay = lrate/epochs\n",
    "sgd = SGD(lr=lrate, momentum=0.9, decay=decay, nesterov=False)\n",
    "model.compile(loss='categorical_crossentropy', optimizer=sgd, metrics=['accuracy'])\n",
    "model.summary()\n",
    "# Fit the model\n",
    "model.fit(X_train, y_train, validation_data=(X_test, y_test), epochs=epochs, batch_size=32)\n",
    "# Final evaluation of the model\n",
    "scores = model.evaluate(X_test, y_test, verbose=0)\n",
    "print(\"Accuracy: %.2f%%\" % (scores[1]*100))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.save('ssd300_mAP_77.43_v2.pth')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "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": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}