location_classify.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "initial_id",
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras import Sequential\n",
    "from tensorflow.keras.layers import Conv2D, MaxPooling2D, Dense, Flatten, Dropout\n",
    "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "2319b913-2cbe-4b05-a48b-1df2ffbffc02",
   "metadata": {},
   "outputs": [],
   "source": [
    "src_path_training = \"geodataset/training/\"\n",
    "src_path_test = \"geodataset/test/\"\n",
    "classes = [\"canada\", \"japan\", \"sweden\", \"united_states\", \"france\"]\n",
    "\n",
    "train_datagen = ImageDataGenerator(\n",
    "    rescale  = 1 / 255.0,\n",
    "    rotation_range = 5,\n",
    "    zoom_range = 0.01,\n",
    "    width_shift_range = 0.05,\n",
    "    height_shift_range = 0.05,\n",
    "    horizontal_flip = True,\n",
    "    fill_mode = \"nearest\"\n",
    ")\n",
    "\n",
    "test_datagen = ImageDataGenerator(rescale = 1 / 255.0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "3d8ee4b8-b955-41a0-8145-024ec4c9a8d1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 500 images belonging to 5 classes.\n",
      "Found 0 images belonging to 5 classes.\n",
      "Found 100 images belonging to 1 classes.\n"
     ]
    }
   ],
   "source": [
    "batch_size = 20\n",
    "train_generator = train_datagen.flow_from_directory(\n",
    "    directory=src_path_training,\n",
    "    target_size=(224, 224),\n",
    "    color_mode=\"rgb\",\n",
    "    batch_size=batch_size,\n",
    "    class_mode=\"categorical\",\n",
    "    subset=\"training\",\n",
    "    shuffle=True,\n",
    "    seed=42\n",
    ")\n",
    "valid_generator = train_datagen.flow_from_directory(\n",
    "    directory=src_path_training,\n",
    "    target_size=(224, 224),\n",
    "    color_mode=\"rgb\",\n",
    "    batch_size=batch_size,\n",
    "    class_mode=\"categorical\",\n",
    "    subset=\"validation\",\n",
    "    shuffle=False,\n",
    "    seed=42\n",
    ")\n",
    "test_generator = test_datagen.flow_from_directory(\n",
    "    directory=src_path_test,\n",
    "    target_size=(224, 224),\n",
    "    color_mode=\"rgb\",\n",
    "    batch_size=1,\n",
    "    class_mode=None,\n",
    "    shuffle=False,\n",
    "    seed=42\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "09a4eb29-d210-40b7-8d92-c43d2789487e",
   "metadata": {},
   "outputs": [],
   "source": [
    "def prepare_model():\n",
    "    model = Sequential()\n",
    "    model.add(Conv2D(32, kernel_size=(3,3), activation='relu', input_shape=(224, 224, 3)))\n",
    "    model.add(MaxPooling2D(pool_size=(2,2)))\n",
    "    model.add(Conv2D(64, kernel_size=(3,3), activation='relu'))\n",
    "    model.add(MaxPooling2D(pool_size=(2,2)))\n",
    "    model.add(Flatten())\n",
    "    model.add(Dense(128, activation='relu'))\n",
    "    model.add(Dense(5, activation='softmax'))\n",
    "    model.compile(loss=\"binary_crossentropy\", optimizer=\"adam\", metrics=['accuracy'])\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "19348886-3dc4-4b42-88e8-f2fd30039f50",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\wesle\\PycharmProjects\\locationclassifier\\venv\\Lib\\site-packages\\keras\\src\\layers\\convolutional\\base_conv.py:107: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.\n",
      "  super().__init__(activity_regularizer=activity_regularizer, **kwargs)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/18\n",
      "\u001b[1m25/25\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m14s\u001b[0m 502ms/step - accuracy: 0.1522 - loss: 1.4144 - val_accuracy: 0.2100 - val_loss: 0.5066\n",
      "Epoch 2/18\n",
      "\u001b[1m25/25\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m4s\u001b[0m 152ms/step - accuracy: 0.0000e+00 - loss: 0.0000e+00 - val_accuracy: 0.2080 - val_loss: 0.5066\n",
      "Epoch 3/18\n",
      "\u001b[1m25/25\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m14s\u001b[0m 511ms/step - accuracy: 0.2661 - loss: 0.4989 - val_accuracy: 0.3340 - val_loss: 0.4743\n",
      "Epoch 4/18\n",
      "\u001b[1m25/25\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m4s\u001b[0m 155ms/step - accuracy: 0.0000e+00 - loss: 0.0000e+00 - val_accuracy: 0.3440 - val_loss: 0.4734\n",
      "Epoch 5/18\n",
      "\u001b[1m25/25\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 491ms/step - accuracy: 0.3261 - loss: 0.4803 - val_accuracy: 0.3400 - val_loss: 0.4797\n",
      "Epoch 6/18\n",
      "\u001b[1m25/25\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m4s\u001b[0m 152ms/step - accuracy: 0.0000e+00 - loss: 0.0000e+00 - val_accuracy: 0.3500 - val_loss: 0.4770\n",
      "Epoch 7/18\n",
      "\u001b[1m25/25\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 486ms/step - accuracy: 0.3512 - loss: 0.4745 - val_accuracy: 0.3820 - val_loss: 0.4539\n",
      "Epoch 8/18\n",
      "\u001b[1m25/25\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m4s\u001b[0m 154ms/step - accuracy: 0.0000e+00 - loss: 0.0000e+00 - val_accuracy: 0.3860 - val_loss: 0.4539\n",
      "Epoch 9/18\n",
      "\u001b[1m25/25\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 486ms/step - accuracy: 0.4220 - loss: 0.4454 - val_accuracy: 0.4680 - val_loss: 0.4401\n",
      "Epoch 10/18\n",
      "\u001b[1m25/25\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m4s\u001b[0m 154ms/step - accuracy: 0.0000e+00 - loss: 0.0000e+00 - val_accuracy: 0.4440 - val_loss: 0.4405\n",
      "Epoch 11/18\n",
      "\u001b[1m25/25\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 480ms/step - accuracy: 0.4524 - loss: 0.4297 - val_accuracy: 0.4140 - val_loss: 0.4339\n",
      "Epoch 12/18\n",
      "\u001b[1m25/25\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m4s\u001b[0m 154ms/step - accuracy: 0.0000e+00 - loss: 0.0000e+00 - val_accuracy: 0.4240 - val_loss: 0.4367\n",
      "Epoch 13/18\n",
      "\u001b[1m25/25\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 485ms/step - accuracy: 0.4350 - loss: 0.4345 - val_accuracy: 0.4580 - val_loss: 0.4184\n",
      "Epoch 14/18\n",
      "\u001b[1m25/25\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m4s\u001b[0m 153ms/step - accuracy: 0.0000e+00 - loss: 0.0000e+00 - val_accuracy: 0.4580 - val_loss: 0.4224\n",
      "Epoch 15/18\n",
      "\u001b[1m25/25\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m14s\u001b[0m 522ms/step - accuracy: 0.5111 - loss: 0.4083 - val_accuracy: 0.4840 - val_loss: 0.4071\n",
      "Epoch 16/18\n",
      "\u001b[1m25/25\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m4s\u001b[0m 163ms/step - accuracy: 0.0000e+00 - loss: 0.0000e+00 - val_accuracy: 0.5140 - val_loss: 0.4032\n",
      "Epoch 17/18\n",
      "\u001b[1m25/25\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m13s\u001b[0m 496ms/step - accuracy: 0.5045 - loss: 0.4021 - val_accuracy: 0.5360 - val_loss: 0.3869\n",
      "Epoch 18/18\n",
      "\u001b[1m25/25\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m4s\u001b[0m 152ms/step - accuracy: 0.0000e+00 - loss: 0.0000e+00 - val_accuracy: 0.5620 - val_loss: 0.3865\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.src.callbacks.history.History at 0x1ba4fbda050>"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = prepare_model()\n",
    "model.fit(train_generator, \n",
    "                    validation_data=train_generator, \n",
    "                    steps_per_epoch=train_generator.n//train_generator.batch_size,\n",
    "                    validation_steps = valid_generator.n//valid_generator.batch_size,\n",
    "                    epochs=18)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "6f93b7b6-0f9d-4f4f-98e3-e311a2a3bd89",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[1m100/100\u001b[0m \u001b[32m━━━━━━━━━━━━━━━━━━━━\u001b[0m\u001b[37m\u001b[0m \u001b[1m1s\u001b[0m 7ms/step\n",
      "for file  predict\\canada_0.jpg  predicted class is  canada  with probability  0.3754546\n",
      "for file  predict\\canada_1.jpg  predicted class is  france  with probability  0.5310502\n",
      "for file  predict\\canada_10.jpg  predicted class is  sweden  with probability  0.7703638\n",
      "for file  predict\\canada_11.jpg  predicted class is  japan  with probability  0.27844375\n",
      "for file  predict\\canada_12.jpg  predicted class is  canada  with probability  0.43088487\n",
      "for file  predict\\canada_13.jpg  predicted class is  canada  with probability  0.38742056\n",
      "for file  predict\\canada_14.jpg  predicted class is  japan  with probability  0.5849925\n",
      "for file  predict\\canada_15.jpg  predicted class is  france  with probability  0.49733913\n",
      "for file  predict\\canada_16.jpg  predicted class is  france  with probability  0.76302373\n",
      "for file  predict\\canada_17.jpg  predicted class is  france  with probability  0.6434376\n",
      "for file  predict\\canada_18.jpg  predicted class is  france  with probability  0.6866854\n",
      "for file  predict\\canada_19.jpg  predicted class is  france  with probability  0.44902077\n",
      "for file  predict\\canada_2.jpg  predicted class is  canada  with probability  0.46752518\n",
      "for file  predict\\canada_3.jpg  predicted class is  france  with probability  0.6061381\n",
      "for file  predict\\canada_4.jpg  predicted class is  sweden  with probability  0.7807341\n",
      "for file  predict\\canada_5.jpg  predicted class is  france  with probability  0.53178966\n",
      "for file  predict\\canada_6.jpg  predicted class is  france  with probability  0.4159978\n",
      "for file  predict\\canada_7.jpg  predicted class is  canada  with probability  0.42502135\n",
      "for file  predict\\canada_8.jpg  predicted class is  japan  with probability  0.36788753\n",
      "for file  predict\\canada_9.jpg  predicted class is  japan  with probability  0.5910506\n",
      "for file  predict\\france_0.jpg  predicted class is  japan  with probability  0.40087372\n",
      "for file  predict\\france_1.jpg  predicted class is  japan  with probability  0.47325802\n",
      "for file  predict\\france_10.jpg  predicted class is  japan  with probability  0.629284\n",
      "for file  predict\\france_11.jpg  predicted class is  japan  with probability  0.42504224\n",
      "for file  predict\\france_12.jpg  predicted class is  japan  with probability  0.6929536\n",
      "for file  predict\\france_13.jpg  predicted class is  japan  with probability  0.88285094\n",
      "for file  predict\\france_14.jpg  predicted class is  japan  with probability  0.35095027\n",
      "for file  predict\\france_15.jpg  predicted class is  japan  with probability  0.5589525\n",
      "for file  predict\\france_16.jpg  predicted class is  sweden  with probability  0.53534114\n",
      "for file  predict\\france_17.jpg  predicted class is  japan  with probability  0.38517517\n",
      "for file  predict\\france_18.jpg  predicted class is  japan  with probability  0.5089375\n",
      "for file  predict\\france_19.jpg  predicted class is  japan  with probability  0.47561753\n",
      "for file  predict\\france_2.jpg  predicted class is  japan  with probability  0.4309371\n",
      "for file  predict\\france_3.jpg  predicted class is  japan  with probability  0.7919133\n",
      "for file  predict\\france_4.jpg  predicted class is  japan  with probability  0.5748663\n",
      "for file  predict\\france_5.jpg  predicted class is  japan  with probability  0.5178628\n",
      "for file  predict\\france_6.jpg  predicted class is  japan  with probability  0.7398079\n",
      "for file  predict\\france_7.jpg  predicted class is  sweden  with probability  0.3467022\n",
      "for file  predict\\france_8.jpg  predicted class is  japan  with probability  0.61276025\n",
      "for file  predict\\france_9.jpg  predicted class is  japan  with probability  0.49508283\n",
      "for file  predict\\japan_0.jpg  predicted class is  japan  with probability  0.30766594\n",
      "for file  predict\\japan_1.jpg  predicted class is  sweden  with probability  0.6979029\n",
      "for file  predict\\japan_10.jpg  predicted class is  france  with probability  0.62573695\n",
      "for file  predict\\japan_11.jpg  predicted class is  japan  with probability  0.39581707\n",
      "for file  predict\\japan_12.jpg  predicted class is  sweden  with probability  0.49651483\n",
      "for file  predict\\japan_13.jpg  predicted class is  sweden  with probability  0.7136446\n",
      "for file  predict\\japan_14.jpg  predicted class is  sweden  with probability  0.62726456\n",
      "for file  predict\\japan_15.jpg  predicted class is  sweden  with probability  0.73826754\n",
      "for file  predict\\japan_16.jpg  predicted class is  sweden  with probability  0.6281415\n",
      "for file  predict\\japan_17.jpg  predicted class is  sweden  with probability  0.73359066\n",
      "for file  predict\\japan_18.jpg  predicted class is  france  with probability  0.4549021\n",
      "for file  predict\\japan_19.jpg  predicted class is  france  with probability  0.5115283\n",
      "for file  predict\\japan_2.jpg  predicted class is  sweden  with probability  0.6826783\n",
      "for file  predict\\japan_3.jpg  predicted class is  japan  with probability  0.6284821\n",
      "for file  predict\\japan_4.jpg  predicted class is  japan  with probability  0.43402913\n",
      "for file  predict\\japan_5.jpg  predicted class is  japan  with probability  0.53383964\n",
      "for file  predict\\japan_6.jpg  predicted class is  france  with probability  0.81658834\n",
      "for file  predict\\japan_7.jpg  predicted class is  sweden  with probability  0.46451923\n",
      "for file  predict\\japan_8.jpg  predicted class is  sweden  with probability  0.3721536\n",
      "for file  predict\\japan_9.jpg  predicted class is  france  with probability  0.41281188\n",
      "for file  predict\\sweden_0.jpg  predicted class is  japan  with probability  0.6484356\n",
      "for file  predict\\sweden_1.jpg  predicted class is  japan  with probability  0.4478939\n",
      "for file  predict\\sweden_10.jpg  predicted class is  france  with probability  0.55072814\n",
      "for file  predict\\sweden_11.jpg  predicted class is  japan  with probability  0.4653884\n",
      "for file  predict\\sweden_12.jpg  predicted class is  united_states  with probability  0.7490501\n",
      "for file  predict\\sweden_13.jpg  predicted class is  united_states  with probability  0.8240386\n",
      "for file  predict\\sweden_14.jpg  predicted class is  france  with probability  0.46866843\n",
      "for file  predict\\sweden_15.jpg  predicted class is  france  with probability  0.46780875\n",
      "for file  predict\\sweden_16.jpg  predicted class is  france  with probability  0.62766904\n",
      "for file  predict\\sweden_17.jpg  predicted class is  united_states  with probability  0.41782886\n",
      "for file  predict\\sweden_18.jpg  predicted class is  france  with probability  0.6766693\n",
      "for file  predict\\sweden_19.jpg  predicted class is  united_states  with probability  0.92030936\n",
      "for file  predict\\sweden_2.jpg  predicted class is  japan  with probability  0.49557325\n",
      "for file  predict\\sweden_3.jpg  predicted class is  japan  with probability  0.5719661\n",
      "for file  predict\\sweden_4.jpg  predicted class is  japan  with probability  0.6209094\n",
      "for file  predict\\sweden_5.jpg  predicted class is  japan  with probability  0.49844897\n",
      "for file  predict\\sweden_6.jpg  predicted class is  japan  with probability  0.44247353\n",
      "for file  predict\\sweden_7.jpg  predicted class is  france  with probability  0.78314364\n",
      "for file  predict\\sweden_8.jpg  predicted class is  united_states  with probability  0.8327706\n",
      "for file  predict\\sweden_9.jpg  predicted class is  japan  with probability  0.44940683\n",
      "for file  predict\\united_states_0.jpg  predicted class is  japan  with probability  0.36620888\n",
      "for file  predict\\united_states_1.jpg  predicted class is  france  with probability  0.7276617\n",
      "for file  predict\\united_states_10.jpg  predicted class is  france  with probability  0.69823766\n",
      "for file  predict\\united_states_11.jpg  predicted class is  france  with probability  0.6821033\n",
      "for file  predict\\united_states_12.jpg  predicted class is  japan  with probability  0.63632745\n",
      "for file  predict\\united_states_13.jpg  predicted class is  france  with probability  0.53868145\n",
      "for file  predict\\united_states_14.jpg  predicted class is  japan  with probability  0.47363457\n",
      "for file  predict\\united_states_15.jpg  predicted class is  france  with probability  0.7629337\n",
      "for file  predict\\united_states_16.jpg  predicted class is  canada  with probability  0.45714155\n",
      "for file  predict\\united_states_17.jpg  predicted class is  france  with probability  0.4983207\n",
      "for file  predict\\united_states_18.jpg  predicted class is  france  with probability  0.54033506\n",
      "for file  predict\\united_states_19.jpg  predicted class is  france  with probability  0.9236195\n",
      "for file  predict\\united_states_2.jpg  predicted class is  canada  with probability  0.7824664\n",
      "for file  predict\\united_states_3.jpg  predicted class is  france  with probability  0.8399084\n",
      "for file  predict\\united_states_4.jpg  predicted class is  france  with probability  0.5605764\n",
      "for file  predict\\united_states_5.jpg  predicted class is  japan  with probability  0.3874472\n",
      "for file  predict\\united_states_6.jpg  predicted class is  france  with probability  0.47361916\n",
      "for file  predict\\united_states_7.jpg  predicted class is  france  with probability  0.71279854\n",
      "for file  predict\\united_states_8.jpg  predicted class is  france  with probability  0.86995596\n",
      "for file  predict\\united_states_9.jpg  predicted class is  france  with probability  0.4558062\n"
     ]
    }
   ],
   "source": [
    "predict = model.predict(test_generator)\n",
    "y_classes = predict.argmax(axis=-1)\n",
    "files = test_generator.filenames\n",
    "for i, p in enumerate(predict):\n",
    "    index=np.argmax(p)\n",
    "    classname = classes[index]\n",
    "    prob = p[index]\n",
    "    print(\"for file \", files[i], ' predicted class is ', classname, ' with probability ', prob)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "44601d50-3e51-488d-a463-456b1bbb37e7",
   "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": "3.11.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}