CNN to classify plastic (if clean or dirty).ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "2a07fcf2-169e-444b-90a2-005bc02840ca",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Librairies\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Conv2D\n",
    "from keras.layers import MaxPooling2D\n",
    "from keras.layers import Flatten\n",
    "from keras.layers import Dense\n",
    "from keras.preprocessing.image import ImageDataGenerator"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "244dffd0-ebbe-45fd-83b0-45c95e200504",
   "metadata": {},
   "outputs": [],
   "source": [
    "#making test model\n",
    "# Initialisation\n",
    "classifier = Sequential()\n",
    "\n",
    "# La Convolution\n",
    "classifier.add(Conv2D(32, (3, 3), input_shape = (64, 64, 3), activation = 'relu'))\n",
    "\n",
    "# Max Pooling\n",
    "classifier.add(MaxPooling2D(pool_size = (2, 2)))\n",
    "\n",
    "# adding conv and maxpooling\n",
    "classifier.add(Conv2D(32, (3, 3), activation = 'relu'))\n",
    "classifier.add(MaxPooling2D(pool_size = (2, 2)))\n",
    "\n",
    "# Flattening\n",
    "classifier.add(Flatten())\n",
    "\n",
    "# connected layer\n",
    "classifier.add(Dense(units = 128, activation = 'relu'))\n",
    "classifier.add(Dense(units = 1, activation = 'sigmoid'))\n",
    "\n",
    "# training parameters\n",
    "classifier.compile(optimizer = 'adam', loss = 'binary_crossentropy', metrics = ['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "id": "4caa0e31-abf2-4c94-a9ca-693f5f9f1f3e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 252 images belonging to 2 classes.\n",
      "Found 92 images belonging to 2 classes.\n",
      "Epoch 1/30\n",
      "8/8 [==============================] - ETA: 0s - loss: 0.4888 - accuracy: 0.7703WARNING:tensorflow:Your input ran out of data; interrupting training. Make sure that your dataset or generator can generate at least `steps_per_epoch * epochs` batches (in this case, 5 batches). You may need to use the repeat() function when building your dataset.\n",
      "8/8 [==============================] - 7s 909ms/step - loss: 0.4888 - accuracy: 0.7703 - val_loss: 0.4197 - val_accuracy: 0.8587\n",
      "Epoch 2/30\n",
      "8/8 [==============================] - 2s 287ms/step - loss: 0.2863 - accuracy: 0.8649\n",
      "Epoch 3/30\n",
      "8/8 [==============================] - 2s 277ms/step - loss: 0.2266 - accuracy: 0.9324\n",
      "Epoch 4/30\n",
      "8/8 [==============================] - 2s 257ms/step - loss: 0.1928 - accuracy: 0.9324\n",
      "Epoch 5/30\n",
      "8/8 [==============================] - 2s 264ms/step - loss: 0.1683 - accuracy: 0.9459\n",
      "Epoch 6/30\n",
      "8/8 [==============================] - 2s 263ms/step - loss: 0.1670 - accuracy: 0.9279\n",
      "Epoch 7/30\n",
      "8/8 [==============================] - 2s 285ms/step - loss: 0.1319 - accuracy: 0.9505\n",
      "Epoch 8/30\n",
      "8/8 [==============================] - 2s 293ms/step - loss: 0.1217 - accuracy: 0.9667\n",
      "Epoch 9/30\n",
      "8/8 [==============================] - 2s 285ms/step - loss: 0.0854 - accuracy: 0.9820\n",
      "Epoch 10/30\n",
      "8/8 [==============================] - 2s 268ms/step - loss: 0.1022 - accuracy: 0.9820\n",
      "Epoch 11/30\n",
      "8/8 [==============================] - 2s 232ms/step - loss: 0.0736 - accuracy: 0.9910\n",
      "Epoch 12/30\n",
      "8/8 [==============================] - 2s 264ms/step - loss: 0.0774 - accuracy: 0.9875\n",
      "Epoch 13/30\n",
      "8/8 [==============================] - 2s 245ms/step - loss: 0.0730 - accuracy: 0.9820\n",
      "Epoch 14/30\n",
      "8/8 [==============================] - 2s 262ms/step - loss: 0.0494 - accuracy: 0.9865\n",
      "Epoch 15/30\n",
      "8/8 [==============================] - 2s 221ms/step - loss: 0.0503 - accuracy: 0.9910\n",
      "Epoch 16/30\n",
      "8/8 [==============================] - 2s 276ms/step - loss: 0.0427 - accuracy: 1.0000\n",
      "Epoch 17/30\n",
      "8/8 [==============================] - 2s 282ms/step - loss: 0.0406 - accuracy: 1.0000\n",
      "Epoch 18/30\n",
      "8/8 [==============================] - 2s 284ms/step - loss: 0.0485 - accuracy: 0.9865\n",
      "Epoch 19/30\n",
      "8/8 [==============================] - 3s 357ms/step - loss: 0.0403 - accuracy: 0.9955\n",
      "Epoch 20/30\n",
      "8/8 [==============================] - 3s 332ms/step - loss: 0.0253 - accuracy: 1.0000\n",
      "Epoch 21/30\n",
      "8/8 [==============================] - 2s 287ms/step - loss: 0.0327 - accuracy: 1.0000\n",
      "Epoch 22/30\n",
      "8/8 [==============================] - 2s 266ms/step - loss: 0.0306 - accuracy: 0.9910\n",
      "Epoch 23/30\n",
      "8/8 [==============================] - 2s 258ms/step - loss: 0.0260 - accuracy: 1.0000\n",
      "Epoch 24/30\n",
      "8/8 [==============================] - 2s 269ms/step - loss: 0.0375 - accuracy: 0.9865\n",
      "Epoch 25/30\n",
      "8/8 [==============================] - 2s 269ms/step - loss: 0.0532 - accuracy: 0.9865\n",
      "Epoch 26/30\n",
      "8/8 [==============================] - 2s 259ms/step - loss: 0.0300 - accuracy: 0.9865\n",
      "Epoch 27/30\n",
      "8/8 [==============================] - 2s 254ms/step - loss: 0.0254 - accuracy: 1.0000\n",
      "Epoch 28/30\n",
      "8/8 [==============================] - 2s 304ms/step - loss: 0.0213 - accuracy: 0.9955\n",
      "Epoch 29/30\n",
      "8/8 [==============================] - 2s 303ms/step - loss: 0.0280 - accuracy: 0.9955\n",
      "Epoch 30/30\n",
      "8/8 [==============================] - 2s 275ms/step - loss: 0.0224 - accuracy: 0.9955\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x16316097610>"
      ]
     },
     "execution_count": 112,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#model to predict if plastic is clean or dirty\n",
    "\n",
    "from keras.applications.vgg16 import VGG16\n",
    "from keras.layers import Dense, Flatten\n",
    "from keras.models import Model\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "\n",
    "# defining the VGG16\n",
    "classifier_vgg16_2 = VGG16(weights='imagenet', include_top=False, input_shape=(64, 64, 3))\n",
    "\n",
    "#Defining the hidden layers as untrainable\n",
    "#(we could retrain them or choose to retrain only the first ones for example)\n",
    "for layer in classifier_vgg16_2.layers:\n",
    "    layer.trainable = False\n",
    "\n",
    "#adding layers specific to our classification problem\n",
    "flat2 = Flatten()(classifier_vgg16_2.output)\n",
    "classif2 = Dense(128, activation='relu', kernel_initializer='he_uniform')(flat2)\n",
    "output2 = Dense(1, activation='sigmoid')(classif2)\n",
    "\n",
    "classifier_vgg16_2 = Model(inputs=classifier_vgg16_2.inputs, outputs=output2)\n",
    "\n",
    "classifier_vgg16_2.compile(optimizer='adam', loss='binary_crossentropy', metrics=['accuracy'])\n",
    "\n",
    "#training\n",
    "train_datagen_2 = ImageDataGenerator(rescale = 1./255,\n",
    "                                   shear_range = 0.2,\n",
    "                                   zoom_range = 0.2,\n",
    "                                   horizontal_flip = True)\n",
    "\n",
    "test_datagen_2 = ImageDataGenerator(rescale = 1./255)\n",
    "\n",
    "training_set_2 = train_datagen_2.flow_from_directory(\"C:/Users/catri/OneDrive/Simplon/IA-P2-Euskadi with my Triof project/Projets/Projet P8 - Triof/triof/cleandirty_training\",\n",
    "                                                 target_size = (64, 64),\n",
    "                                                 batch_size = 30,\n",
    "                                                 class_mode = 'binary')\n",
    "\n",
    "test_set_2 = test_datagen_2.flow_from_directory(\"C:/Users/catri/OneDrive/Simplon/IA-P2-Euskadi with my Triof project/Projets/Projet P8 - Triof/triof/cleandirty_test\",\n",
    "                                            target_size = (64, 64),\n",
    "                                            batch_size = 30,\n",
    "                                            class_mode = 'binary')\n",
    "\n",
    "classifier_vgg16_2.fit(training_set_2,\n",
    "               steps_per_epoch = 8, #250/30 = 8.33\n",
    "               epochs = 30,\n",
    "               validation_data = test_set_2,\n",
    "               validation_steps=5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 121,
   "id": "4a46787e-533d-4341-8068-d569271d6c71",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[1.]]\n",
      "Item is dirty\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": [
    "from keras.preprocessing import image\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "#showing test image\n",
    "img = image.load_img(\"C:/Users/catri/OneDrive/Simplon/IA-P2-Euskadi with my Triof project/Projets/Projet P8 - Triof/triof/cleandirty_test/dirty_plastics/0f8ada3d14.jpg\",target_size=(64,64))\n",
    "img = np.asarray(img)\n",
    "plt.imshow(img)\n",
    "img = np.expand_dims(img, axis=0)\n",
    "\n",
    "#making prediction\n",
    "output = classifier_vgg16_2.predict(img)\n",
    "print(output)\n",
    "\n",
    "#returning prediction\n",
    "if output <= 0.5:\n",
    "    print('Item is clean')\n",
    "else:\n",
    "    print('Item is dirty')\n",
    "    \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 122,
   "id": "0e2f1e34-5b0d-42a9-ab6c-247cc0014562",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From C:\\Users\\catri\\anaconda3\\lib\\site-packages\\tensorflow\\python\\training\\tracking\\tracking.py:111: Model.state_updates (from tensorflow.python.keras.engine.training) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "This property should not be used in TensorFlow 2.0, as updates are applied automatically.\n",
      "WARNING:tensorflow:From C:\\Users\\catri\\anaconda3\\lib\\site-packages\\tensorflow\\python\\training\\tracking\\tracking.py:111: Layer.updates (from tensorflow.python.keras.engine.base_layer) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "This property should not be used in TensorFlow 2.0, as updates are applied automatically.\n",
      "INFO:tensorflow:Assets written to: clean or dirty classifier\\assets\n"
     ]
    }
   ],
   "source": [
    "#saving the model\n",
    "classifier_vgg16_2.save(\"clean or dirty classifier\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "id": "b54389cd-9bd1-4ad0-943e-9296f3b62f4e",
   "metadata": {},
   "outputs": [],
   "source": [
    "#if want to load it use:\n",
    "\n",
    "#from keras.models import load_model\n",
    "#classifier = load_model('clean or dirty classifier')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3c77d82e-d48d-4912-95d7-00e1ff5f2636",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.8.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}