gold_vgg.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "736c4750",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np \n",
    "from tqdm import tqdm\n",
    "import cv2\n",
    "import os\n",
    "import shutil\n",
    "import itertools\n",
    "#import imutils\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.preprocessing import LabelBinarizer\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import accuracy_score, confusion_matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "900dd824",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "x = './data/train/malignant'\n",
    "y  = './data/train/benign'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "16ff5f36",
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import os\n",
    "import PIL\n",
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "from tensorflow.keras import layers\n",
    "from tensorflow.python.keras.layers import Dense, Flatten\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.optimizers import Adam\n",
    "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
    "\n",
    "from glob import glob\n",
    "from tensorflow.keras import backend as K\n",
    "import gc\n",
    "\n",
    "from tensorflow.keras.callbacks import Callback, ModelCheckpoint, ReduceLROnPlateau, TensorBoard\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import cohen_kappa_score, accuracy_score\n",
    "import scipy\n",
    "from tqdm import tqdm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "9cef793c",
   "metadata": {},
   "outputs": [],
   "source": [
    "SIZE=256\n",
    "base_model = keras.applications.vgg19.VGG19(input_shape=(SIZE,SIZE,3),include_top=False,weights='imagenet')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "5da85c48",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"vgg19\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         [(None, 256, 256, 3)]     0         \n",
      "_________________________________________________________________\n",
      "block1_conv1 (Conv2D)        (None, 256, 256, 64)      1792      \n",
      "_________________________________________________________________\n",
      "block1_conv2 (Conv2D)        (None, 256, 256, 64)      36928     \n",
      "_________________________________________________________________\n",
      "block1_pool (MaxPooling2D)   (None, 128, 128, 64)      0         \n",
      "_________________________________________________________________\n",
      "block2_conv1 (Conv2D)        (None, 128, 128, 128)     73856     \n",
      "_________________________________________________________________\n",
      "block2_conv2 (Conv2D)        (None, 128, 128, 128)     147584    \n",
      "_________________________________________________________________\n",
      "block2_pool (MaxPooling2D)   (None, 64, 64, 128)       0         \n",
      "_________________________________________________________________\n",
      "block3_conv1 (Conv2D)        (None, 64, 64, 256)       295168    \n",
      "_________________________________________________________________\n",
      "block3_conv2 (Conv2D)        (None, 64, 64, 256)       590080    \n",
      "_________________________________________________________________\n",
      "block3_conv3 (Conv2D)        (None, 64, 64, 256)       590080    \n",
      "_________________________________________________________________\n",
      "block3_conv4 (Conv2D)        (None, 64, 64, 256)       590080    \n",
      "_________________________________________________________________\n",
      "block3_pool (MaxPooling2D)   (None, 32, 32, 256)       0         \n",
      "_________________________________________________________________\n",
      "block4_conv1 (Conv2D)        (None, 32, 32, 512)       1180160   \n",
      "_________________________________________________________________\n",
      "block4_conv2 (Conv2D)        (None, 32, 32, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block4_conv3 (Conv2D)        (None, 32, 32, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block4_conv4 (Conv2D)        (None, 32, 32, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block4_pool (MaxPooling2D)   (None, 16, 16, 512)       0         \n",
      "_________________________________________________________________\n",
      "block5_conv1 (Conv2D)        (None, 16, 16, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv2 (Conv2D)        (None, 16, 16, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv3 (Conv2D)        (None, 16, 16, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv4 (Conv2D)        (None, 16, 16, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block5_pool (MaxPooling2D)   (None, 8, 8, 512)         0         \n",
      "=================================================================\n",
      "Total params: 20,024,384\n",
      "Trainable params: 20,024,384\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "base_model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "6de7f2cb",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"vgg19\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         [(None, 256, 256, 3)]     0         \n",
      "_________________________________________________________________\n",
      "block1_conv1 (Conv2D)        (None, 256, 256, 64)      1792      \n",
      "_________________________________________________________________\n",
      "block1_conv2 (Conv2D)        (None, 256, 256, 64)      36928     \n",
      "_________________________________________________________________\n",
      "block1_pool (MaxPooling2D)   (None, 128, 128, 64)      0         \n",
      "_________________________________________________________________\n",
      "block2_conv1 (Conv2D)        (None, 128, 128, 128)     73856     \n",
      "_________________________________________________________________\n",
      "block2_conv2 (Conv2D)        (None, 128, 128, 128)     147584    \n",
      "_________________________________________________________________\n",
      "block2_pool (MaxPooling2D)   (None, 64, 64, 128)       0         \n",
      "_________________________________________________________________\n",
      "block3_conv1 (Conv2D)        (None, 64, 64, 256)       295168    \n",
      "_________________________________________________________________\n",
      "block3_conv2 (Conv2D)        (None, 64, 64, 256)       590080    \n",
      "_________________________________________________________________\n",
      "block3_conv3 (Conv2D)        (None, 64, 64, 256)       590080    \n",
      "_________________________________________________________________\n",
      "block3_conv4 (Conv2D)        (None, 64, 64, 256)       590080    \n",
      "_________________________________________________________________\n",
      "block3_pool (MaxPooling2D)   (None, 32, 32, 256)       0         \n",
      "_________________________________________________________________\n",
      "block4_conv1 (Conv2D)        (None, 32, 32, 512)       1180160   \n",
      "_________________________________________________________________\n",
      "block4_conv2 (Conv2D)        (None, 32, 32, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block4_conv3 (Conv2D)        (None, 32, 32, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block4_conv4 (Conv2D)        (None, 32, 32, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block4_pool (MaxPooling2D)   (None, 16, 16, 512)       0         \n",
      "_________________________________________________________________\n",
      "block5_conv1 (Conv2D)        (None, 16, 16, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv2 (Conv2D)        (None, 16, 16, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv3 (Conv2D)        (None, 16, 16, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv4 (Conv2D)        (None, 16, 16, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "block5_pool (MaxPooling2D)   (None, 8, 8, 512)         0         \n",
      "=================================================================\n",
      "Total params: 20,024,384\n",
      "Trainable params: 0\n",
      "Non-trainable params: 20,024,384\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "base_model.trainable = False\n",
    "base_model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "e422977c",
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow.keras import regularizers\n",
    "\n",
    "inputs = keras.Input(shape=(SIZE,SIZE,3))\n",
    "x = keras.applications.vgg19.preprocess_input(inputs)\n",
    "x = base_model(x, training=False)\n",
    "x = layers.GlobalAveragePooling2D()(x)\n",
    "x = layers.Flatten()(x)\n",
    "x = layers.Dense(128, activation='relu',kernel_regularizer=regularizers.l2(0.003))(x)\n",
    "x = layers.Dropout(0.35)(x)\n",
    "x = layers.Dense(64, activation='relu',kernel_regularizer=regularizers.l2(0.003))(x)\n",
    "x = layers.Dropout(0.35)(x)\n",
    "outputs = layers.Dense(1, activation='sigmoid')(x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "75807b77",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"functional_1\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_2 (InputLayer)         [(None, 256, 256, 3)]     0         \n",
      "_________________________________________________________________\n",
      "tf_op_layer_strided_slice (T [(None, 256, 256, 3)]     0         \n",
      "_________________________________________________________________\n",
      "tf_op_layer_BiasAdd (TensorF [(None, 256, 256, 3)]     0         \n",
      "_________________________________________________________________\n",
      "vgg19 (Functional)           (None, 8, 8, 512)         20024384  \n",
      "_________________________________________________________________\n",
      "global_average_pooling2d (Gl (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 128)               65664     \n",
      "_________________________________________________________________\n",
      "dropout (Dropout)            (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 64)                8256      \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 64)                0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 1)                 65        \n",
      "=================================================================\n",
      "Total params: 20,098,369\n",
      "Trainable params: 73,985\n",
      "Non-trainable params: 20,024,384\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model = keras.Model(inputs, outputs)\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "28835bf0",
   "metadata": {},
   "outputs": [],
   "source": [
    "diretorio = './data/train/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "d5716cf9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 2637 files belonging to 2 classes.\n",
      "Using 2110 files for training.\n",
      "Found 2637 files belonging to 2 classes.\n",
      "Using 527 files for validation.\n"
     ]
    }
   ],
   "source": [
    "image_size = (SIZE, SIZE)\n",
    "batch_size = 32\n",
    "\n",
    "train_ds = tf.keras.preprocessing.image_dataset_from_directory(\n",
    "    diretorio,\n",
    "    validation_split=0.2,\n",
    "    subset=\"training\",\n",
    "    seed=1337,\n",
    "    image_size=image_size,\n",
    "    batch_size=batch_size,\n",
    ")\n",
    "val_ds = tf.keras.preprocessing.image_dataset_from_directory(\n",
    "    diretorio,\n",
    "    validation_split=0.2,\n",
    "    subset=\"validation\",\n",
    "    seed=1337,\n",
    "    image_size=image_size,\n",
    "    batch_size=batch_size,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "95987094",
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(\n",
    "    optimizer=keras.optimizers.Adam(1e-3),\n",
    "    loss=\"binary_crossentropy\",\n",
    "    metrics=[\"accuracy\"],\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "6a61d88b",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/30\n",
      "66/66 [==============================] - 289s 4s/step - loss: 1.3188 - accuracy: 0.7318 - val_loss: 0.9993 - val_accuracy: 0.8368\n",
      "Epoch 2/30\n",
      "66/66 [==============================] - 249s 4s/step - loss: 0.9612 - accuracy: 0.8209 - val_loss: 0.8429 - val_accuracy: 0.8520\n",
      "Epoch 3/30\n",
      "66/66 [==============================] - 235s 4s/step - loss: 0.8193 - accuracy: 0.8336 - val_loss: 0.7378 - val_accuracy: 0.8520\n",
      "Epoch 4/30\n",
      "66/66 [==============================] - 232s 4s/step - loss: 0.7353 - accuracy: 0.8408 - val_loss: 0.6703 - val_accuracy: 0.8710\n",
      "Epoch 5/30\n",
      "66/66 [==============================] - 232s 4s/step - loss: 0.6417 - accuracy: 0.8588 - val_loss: 0.6106 - val_accuracy: 0.8786\n",
      "Epoch 6/30\n",
      "66/66 [==============================] - 232s 4s/step - loss: 0.6039 - accuracy: 0.8602 - val_loss: 0.5818 - val_accuracy: 0.8577\n",
      "Epoch 7/30\n",
      "66/66 [==============================] - 245s 4s/step - loss: 0.5412 - accuracy: 0.8711 - val_loss: 0.5452 - val_accuracy: 0.8653\n",
      "Epoch 8/30\n",
      "66/66 [==============================] - 249s 4s/step - loss: 0.5114 - accuracy: 0.8720 - val_loss: 0.5365 - val_accuracy: 0.8463\n",
      "Epoch 9/30\n",
      "66/66 [==============================] - 250s 4s/step - loss: 0.4840 - accuracy: 0.8782 - val_loss: 0.4989 - val_accuracy: 0.8691\n",
      "Epoch 10/30\n",
      "66/66 [==============================] - 246s 4s/step - loss: 0.4497 - accuracy: 0.8839 - val_loss: 0.4873 - val_accuracy: 0.8748\n",
      "Epoch 11/30\n",
      "66/66 [==============================] - 252s 4s/step - loss: 0.4429 - accuracy: 0.8858 - val_loss: 0.4686 - val_accuracy: 0.8861\n",
      "Epoch 12/30\n",
      "66/66 [==============================] - 246s 4s/step - loss: 0.4119 - accuracy: 0.8900 - val_loss: 0.4746 - val_accuracy: 0.8805\n",
      "Epoch 13/30\n",
      "66/66 [==============================] - 325s 5s/step - loss: 0.3889 - accuracy: 0.9019 - val_loss: 0.4521 - val_accuracy: 0.8861\n",
      "Epoch 14/30\n",
      "66/66 [==============================] - 371s 6s/step - loss: 0.3725 - accuracy: 0.8967 - val_loss: 0.4350 - val_accuracy: 0.8634\n",
      "Epoch 15/30\n",
      "66/66 [==============================] - 246s 4s/step - loss: 0.3603 - accuracy: 0.9076 - val_loss: 0.4337 - val_accuracy: 0.8729\n",
      "Epoch 16/30\n",
      "66/66 [==============================] - 247s 4s/step - loss: 0.3478 - accuracy: 0.9052 - val_loss: 0.4359 - val_accuracy: 0.8558\n",
      "Epoch 17/30\n",
      "66/66 [==============================] - 244s 4s/step - loss: 0.3452 - accuracy: 0.9047 - val_loss: 0.4415 - val_accuracy: 0.8520\n",
      "Epoch 18/30\n",
      "66/66 [==============================] - 249s 4s/step - loss: 0.3223 - accuracy: 0.9204 - val_loss: 0.4451 - val_accuracy: 0.8748\n",
      "Epoch 19/30\n",
      "66/66 [==============================] - 237s 4s/step - loss: 0.3163 - accuracy: 0.9147 - val_loss: 0.4458 - val_accuracy: 0.8748\n",
      "Epoch 20/30\n",
      "66/66 [==============================] - 232s 4s/step - loss: 0.3086 - accuracy: 0.9194 - val_loss: 0.4301 - val_accuracy: 0.8880\n",
      "Epoch 21/30\n",
      "66/66 [==============================] - 237s 4s/step - loss: 0.2998 - accuracy: 0.9180 - val_loss: 0.4325 - val_accuracy: 0.8710\n",
      "Epoch 22/30\n",
      "66/66 [==============================] - 232s 4s/step - loss: 0.2935 - accuracy: 0.9265 - val_loss: 0.4206 - val_accuracy: 0.8729\n",
      "Epoch 23/30\n",
      "66/66 [==============================] - 233s 4s/step - loss: 0.2880 - accuracy: 0.9256 - val_loss: 0.4525 - val_accuracy: 0.8596\n",
      "Epoch 24/30\n",
      "66/66 [==============================] - 230s 3s/step - loss: 0.2972 - accuracy: 0.9185 - val_loss: 0.4269 - val_accuracy: 0.8805\n",
      "Epoch 25/30\n",
      "66/66 [==============================] - 240s 4s/step - loss: 0.2854 - accuracy: 0.9223 - val_loss: 0.4326 - val_accuracy: 0.8710\n",
      "Epoch 26/30\n",
      "66/66 [==============================] - 227s 3s/step - loss: 0.2700 - accuracy: 0.9265 - val_loss: 0.4483 - val_accuracy: 0.8767\n",
      "Epoch 27/30\n",
      "66/66 [==============================] - 230s 3s/step - loss: 0.2687 - accuracy: 0.9308 - val_loss: 0.4382 - val_accuracy: 0.8653\n",
      "Epoch 28/30\n",
      "66/66 [==============================] - 234s 4s/step - loss: 0.2539 - accuracy: 0.9341 - val_loss: 0.4497 - val_accuracy: 0.8691\n",
      "Epoch 29/30\n",
      "66/66 [==============================] - 244s 4s/step - loss: 0.2774 - accuracy: 0.9270 - val_loss: 0.4269 - val_accuracy: 0.8653\n",
      "Epoch 30/30\n",
      "66/66 [==============================] - 242s 4s/step - loss: 0.2573 - accuracy: 0.9318 - val_loss: 0.4489 - val_accuracy: 0.8691\n"
     ]
    }
   ],
   "source": [
    "epochs = 30\n",
    "history = model.fit(\n",
    "    train_ds, epochs=epochs, validation_data=val_ds,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "704e2b7c",
   "metadata": {},
   "outputs": [
    {
     "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": [
    "fig1 = plt.gcf()\n",
    "plt.plot(history.history['accuracy'])\n",
    "plt.plot(history.history['val_accuracy'])\n",
    "plt.axis(ymin=0.4,ymax=1)\n",
    "plt.grid()\n",
    "plt.title('Model Accuracy')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.xlabel('Epochs')\n",
    "plt.legend(['train', 'validation'])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "855dc9cd",
   "metadata": {},
   "outputs": [
    {
     "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": [
    "plt.plot(history.history['loss'])\n",
    "plt.plot(history.history['val_loss'])\n",
    "plt.grid()\n",
    "plt.title('Model Loss')\n",
    "plt.ylabel('Loss')\n",
    "plt.xlabel('Epochs')\n",
    "plt.legend(['train', 'validation'])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 141,
   "id": "7dc2b3cb",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1, 256, 256, 3)\n"
     ]
    }
   ],
   "source": [
    "import cv2\n",
    "#inp = r\"/content/drive/MyDrive/code/data/test/melanoma/ISIC_0014766.jpg\"\n",
    "#inp = \"../../code/data/train/melanoma/ISIC_0000029.jpg\"\n",
    "inp = r\"./test/benign/1074.jpg\"\n",
    "\n",
    "\n",
    "image=cv2.imread(inp)\n",
    "image_resized= cv2.resize(image, (256,256))\n",
    "image=np.expand_dims(image_resized,axis=0)\n",
    "print(image.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 142,
   "id": "64030309",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[0.09033343]]\n"
     ]
    }
   ],
   "source": [
    "CATEGORIES = [\"BENIGN\",\"MALIGNANT\"]\n",
    "pred=model.predict(image)\n",
    "print(pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 143,
   "id": "63b34c28",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 143,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "def find_class(result):\n",
    "    if result >= 0.5:\n",
    "        return 1\n",
    "    else:\n",
    "        return 0\n",
    "\n",
    "find_class(pred[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 144,
   "id": "11865c26",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The predicted class is BENIGN\n"
     ]
    }
   ],
   "source": [
    "output_class=CATEGORIES[find_class(pred[0])]\n",
    "print(\"The predicted class is\", output_class)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d7869fdd",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 158,
   "id": "c801cf20",
   "metadata": {},
   "outputs": [],
   "source": [
    "x_test=[]\n",
    "test_path = \"./test\"\n",
    "\n",
    "for folder in os.listdir(test_path):\n",
    "\n",
    "    sub_path=test_path+\"/\"+folder\n",
    "\n",
    "    for img in os.listdir(sub_path):\n",
    "\n",
    "        image_path=sub_path+\"/\"+img\n",
    "\n",
    "        img_arr=cv2.imread(image_path)\n",
    "\n",
    "        img_arr=cv2.resize(img_arr,(224,224))\n",
    "\n",
    "        x_test.append(img_arr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 159,
   "id": "a933b680",
   "metadata": {},
   "outputs": [],
   "source": [
    "test_datagen = ImageDataGenerator(rescale = 1./255)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 160,
   "id": "083bdb7c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 660 images belonging to 2 classes.\n"
     ]
    }
   ],
   "source": [
    "test_set = test_datagen.flow_from_directory(test_path,\n",
    "                                            target_size = (224, 224),\n",
    "                                            batch_size = 32,\n",
    "                                            class_mode = 'sparse')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "85c21960",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "474668de",
   "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": "3.8.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}