Butterfly_Classification.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 346
    },
    "id": "WQe8UU4FA1dU",
    "outputId": "6c42bd07-eebe-4848-ea6b-883a6560196b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Collecting efficientnet\n",
      "  Using cached efficientnet-1.1.1-py3-none-any.whl (18 kB)\n",
      "Collecting keras-applications<=1.0.8,>=1.0.7\n",
      "  Using cached Keras_Applications-1.0.8-py3-none-any.whl (50 kB)\n",
      "Collecting scikit-image\n",
      "  Downloading scikit_image-0.18.0-cp38-cp38-manylinux1_x86_64.whl (30.2 MB)\n",
      "\u001b[K     |████████████████████████████████| 30.2 MB 10.4 MB/s eta 0:00:01\n",
      "\u001b[?25hCollecting h5py\n",
      "  Downloading h5py-3.1.0-cp38-cp38-manylinux1_x86_64.whl (4.4 MB)\n",
      "\u001b[K     |████████████████████████████████| 4.4 MB 8.0 MB/s eta 0:00:01\n",
      "\u001b[?25hRequirement already satisfied: numpy>=1.9.1 in /home/prateek/vpy/lib/python3.8/site-packages (from keras-applications<=1.0.8,>=1.0.7->efficientnet) (1.19.1)\n",
      "Requirement already satisfied: scipy>=1.0.1 in /home/prateek/vpy/lib/python3.8/site-packages (from scikit-image->efficientnet) (1.5.2)\n",
      "Requirement already satisfied: matplotlib!=3.0.0,>=2.0.0 in /home/prateek/vpy/lib/python3.8/site-packages (from scikit-image->efficientnet) (3.3.1)\n",
      "Requirement already satisfied: pillow!=7.1.0,!=7.1.1,>=4.3.0 in /home/prateek/vpy/lib/python3.8/site-packages (from scikit-image->efficientnet) (6.2.1)\n",
      "Requirement already satisfied: networkx>=2.0 in /home/prateek/vpy/lib/python3.8/site-packages (from scikit-image->efficientnet) (2.5)\n",
      "Collecting imageio>=2.3.0\n",
      "  Downloading imageio-2.9.0-py3-none-any.whl (3.3 MB)\n",
      "\u001b[K     |████████████████████████████████| 3.3 MB 11.8 MB/s eta 0:00:01\n",
      "\u001b[?25hCollecting PyWavelets>=1.1.1\n",
      "  Downloading PyWavelets-1.1.1-cp38-cp38-manylinux1_x86_64.whl (4.4 MB)\n",
      "\u001b[K     |████████████████████████████████| 4.4 MB 10.4 MB/s eta 0:00:01\n",
      "\u001b[?25hCollecting tifffile>=2019.7.26\n",
      "  Downloading tifffile-2020.12.8-py3-none-any.whl (157 kB)\n",
      "\u001b[K     |████████████████████████████████| 157 kB 12.7 MB/s eta 0:00:01\n",
      "\u001b[?25hRequirement already satisfied: certifi>=2020.06.20 in /home/prateek/vpy/lib/python3.8/site-packages (from matplotlib!=3.0.0,>=2.0.0->scikit-image->efficientnet) (2020.6.20)\n",
      "Requirement already satisfied: pyparsing!=2.0.4,!=2.1.2,!=2.1.6,>=2.0.3 in /home/prateek/vpy/lib/python3.8/site-packages (from matplotlib!=3.0.0,>=2.0.0->scikit-image->efficientnet) (2.4.7)\n",
      "Requirement already satisfied: cycler>=0.10 in /home/prateek/vpy/lib/python3.8/site-packages (from matplotlib!=3.0.0,>=2.0.0->scikit-image->efficientnet) (0.10.0)\n",
      "Requirement already satisfied: python-dateutil>=2.1 in /home/prateek/vpy/lib/python3.8/site-packages (from matplotlib!=3.0.0,>=2.0.0->scikit-image->efficientnet) (2.8.1)\n",
      "Requirement already satisfied: kiwisolver>=1.0.1 in /home/prateek/vpy/lib/python3.8/site-packages (from matplotlib!=3.0.0,>=2.0.0->scikit-image->efficientnet) (1.2.0)\n",
      "Requirement already satisfied: decorator>=4.3.0 in /home/prateek/vpy/lib/python3.8/site-packages (from networkx>=2.0->scikit-image->efficientnet) (4.4.2)\n",
      "Requirement already satisfied: six in /home/prateek/vpy/lib/python3.8/site-packages (from cycler>=0.10->matplotlib!=3.0.0,>=2.0.0->scikit-image->efficientnet) (1.15.0)\n",
      "Installing collected packages: h5py, keras-applications, imageio, PyWavelets, tifffile, scikit-image, efficientnet\n",
      "Successfully installed PyWavelets-1.1.1 efficientnet-1.1.1 h5py-3.1.0 imageio-2.9.0 keras-applications-1.0.8 scikit-image-0.18.0 tifffile-2020.12.8\n",
      "\u001b[33mWARNING: You are using pip version 20.2.3; however, version 20.3.1 is available.\n",
      "You should consider upgrading via the '/home/prateek/vpy/bin/python3 -m pip install --upgrade pip' command.\u001b[0m\n"
     ]
    }
   ],
   "source": [
    "!pip install efficientnet"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "id": "zAnqTupPAdp4"
   },
   "outputs": [
    {
     "ename": "ModuleNotFoundError",
     "evalue": "No module named 'tensorflow'",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mModuleNotFoundError\u001b[0m                       Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-2-603674d4cd81>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0msklearn\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmodel_selection\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mStratifiedKFold\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mtqdm\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 5\u001b[0;31m \u001b[0;32mimport\u001b[0m \u001b[0mtensorflow\u001b[0m \u001b[0;32mas\u001b[0m \u001b[0mtf\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      6\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0mtensorflow\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mkeras\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpreprocessing\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mimage\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mImageDataGenerator\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      7\u001b[0m \u001b[0;32mfrom\u001b[0m \u001b[0mtensorflow\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mkeras\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mapplications\u001b[0m \u001b[0;32mimport\u001b[0m \u001b[0mResNet50\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mModuleNotFoundError\u001b[0m: No module named 'tensorflow'"
     ]
    }
   ],
   "source": [
    "import pandas as pd \n",
    "import numpy as np \n",
    "from sklearn.model_selection import StratifiedKFold\n",
    "import tqdm\n",
    "import tensorflow as tf\n",
    "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
    "from tensorflow.keras.applications import ResNet50\n",
    "from tensorflow.keras.layers import Dense,Dropout,BatchNormalization,GlobalAveragePooling2D\n",
    "from tensorflow.keras.models import Sequential,load_model\n",
    "from tensorflow.keras.optimizers import Adam\n",
    "import tensorflow_addons as tfa \n",
    "from tensorflow.keras.callbacks import ReduceLROnPlateau,ModelCheckpoint,EarlyStopping\n",
    "import matplotlib.pyplot as plt\n",
    "import tensorflow.keras.backend as K\n",
    "import keras \n",
    "from keras.utils import to_categorical\n",
    "from sklearn.metrics import confusion_matrix\n",
    "import efficientnet.tfkeras as efn "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "id": "pfe1aK-3Aqqo",
    "outputId": "aaaffd70-24d2-4ff8-f205-92c2629b9e46"
   },
   "outputs": [],
   "source": [
    "# !wget -O \"butterfly_classification_ai_challenge-dataset.zip\" \"https://dockship-job-models.s3.ap-south-1.amazonaws.com/d937cfa8a2379f55fb8a06fe24ec61ac?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIAIDOPTEUZ2LEOQEGQ%2F20201019%2Fap-south-1%2Fs3%2Faws4_request&X-Amz-Date=20201019T121320Z&X-Amz-Expires=1800&X-Amz-Signature=1736f50e51c72d51d96aa818697b70fd06062db98e129a84c48244f6c5c5a304&X-Amz-SignedHeaders=host&response-content-disposition=attachment%3B%20filename%3D%22butterfly_classification_ai_challenge-dataset.zip%22\"\n",
    "# !unzip 'butterfly_classification_ai_challenge-dataset.zip' -d 'data'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "id": "a-hvI5hWAdp9"
   },
   "outputs": [],
   "source": [
    "SEED=42\n",
    "TRAIN_DIR='data/DATA/TRAIN'\n",
    "TEST_DIR='data/DATA/TEST'\n",
    "IMG_SIZE=(512,512)\n",
    "BATCHSIZE=16\n",
    "INPUT_SHAPE=(512,512,3)\n",
    "EPOCHS=70\n",
    "LR=1e-4"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "id": "feSjWZ9OAdqA"
   },
   "outputs": [],
   "source": [
    "train_datagen=ImageDataGenerator(rescale=1./512,validation_split=0.1,rotation_range=40,\n",
    "      width_shift_range=0.2,\n",
    "      height_shift_range=0.2,\n",
    "      shear_range=0.2,\n",
    "      zoom_range=0.2,\n",
    "      horizontal_flip=True,\n",
    "      fill_mode='nearest')\n",
    "valid_datagen=ImageDataGenerator(rescale=1./512,validation_split=0.9)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 35
    },
    "id": "7_loaMgtAdqE",
    "outputId": "bda2903a-1264-4cfc-e265-39f09f8ff8ee"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 4050 images belonging to 50 classes.\n"
     ]
    }
   ],
   "source": [
    "train_batches=train_datagen.flow_from_directory(TRAIN_DIR,target_size=IMG_SIZE,batch_size=BATCHSIZE,shuffle=True,seed=SEED,subset='training')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 35
    },
    "id": "nFmJW-DjAdqI",
    "outputId": "709d0342-7cad-4ac2-e640-cbd8b38ddd75"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 429 images belonging to 50 classes.\n"
     ]
    }
   ],
   "source": [
    "valid_batches=train_datagen.flow_from_directory(TRAIN_DIR,target_size=IMG_SIZE,batch_size=BATCHSIZE,shuffle=False,seed=SEED,subset='validation')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "id": "3FZ2F5lBAdqM"
   },
   "outputs": [],
   "source": [
    "res=ResNet50(weights='imagenet',include_top=False,input_shape=INPUT_SHAPE)\n",
    "eff=efn.EfficientNetB0(weights='imagenet',include_top=False,input_shape=INPUT_SHAPE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "id": "1Hykkko9AdqP"
   },
   "outputs": [],
   "source": [
    "def f1_metric(y_true, y_pred):\n",
    "    true_positives = K.sum(K.round(K.clip(y_true * y_pred, 0, 1)))\n",
    "    possible_positives = K.sum(K.round(K.clip(y_true, 0, 1)))\n",
    "    predicted_positives = K.sum(K.round(K.clip(y_pred, 0, 1)))\n",
    "    precision = true_positives / (predicted_positives + K.epsilon())\n",
    "    recall = true_positives / (possible_positives + K.epsilon())\n",
    "    f1_val = 2*(precision*recall)/(precision+recall+K.epsilon())\n",
    "    return f1_val\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "id": "bJ7P4jfTAdqT"
   },
   "outputs": [],
   "source": [
    "model=Sequential()\n",
    "model.add(eff)\n",
    "model.add(Dropout(0.5))\n",
    "model.add(GlobalAveragePooling2D())\n",
    "model.add(Dense(50,activation='softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "id": "ydzv8vFAAdqW"
   },
   "outputs": [],
   "source": [
    "model.compile(optimizer=Adam(lr=LR),loss='categorical_crossentropy',metrics=[f1_metric])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "id": "gYHMNzlLAdqZ"
   },
   "outputs": [],
   "source": [
    "redlr=ReduceLROnPlateau(monitor='val_loss',patience=3,verbose=1)\n",
    "chkpt=ModelCheckpoint('best.h5',verbose=1,monitor='val_loss',save_best_only=True)\n",
    "es=EarlyStopping(patience=8,verbose=1,restore_best_weights=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 310
    },
    "id": "_DbH1z8oAdqb",
    "outputId": "6ac453b3-d4ec-4eeb-c1cf-e56c1a746932"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/70\n",
      "  2/254 [..............................] - ETA: 1:40 - loss: 3.9228 - f1_metric: 0.0000e+00WARNING:tensorflow:Callbacks method `on_train_batch_end` is slow compared to the batch time (batch time: 0.2609s vs `on_train_batch_end` time: 0.5345s). Check your callbacks.\n",
      "254/254 [==============================] - ETA: 0s - loss: 2.6591 - f1_metric: 0.1257\n",
      "Epoch 00001: val_loss improved from inf to 0.96895, saving model to best.h5\n",
      "254/254 [==============================] - 358s 1s/step - loss: 2.6591 - f1_metric: 0.1257 - val_loss: 0.9689 - val_f1_metric: 0.6138\n",
      "Epoch 2/70\n",
      "254/254 [==============================] - ETA: 0s - loss: 0.7485 - f1_metric: 0.7535\n",
      "Epoch 00002: val_loss improved from 0.96895 to 0.35072, saving model to best.h5\n",
      "254/254 [==============================] - 356s 1s/step - loss: 0.7485 - f1_metric: 0.7535 - val_loss: 0.3507 - val_f1_metric: 0.9058\n",
      "Epoch 3/70\n",
      "254/254 [==============================] - ETA: 0s - loss: 0.3873 - f1_metric: 0.8944\n",
      "Epoch 00003: val_loss improved from 0.35072 to 0.24477, saving model to best.h5\n",
      "254/254 [==============================] - 357s 1s/step - loss: 0.3873 - f1_metric: 0.8944 - val_loss: 0.2448 - val_f1_metric: 0.9246\n",
      "Epoch 4/70\n",
      "218/254 [========================>.....] - ETA: 47s - loss: 0.2937 - f1_metric: 0.9171"
     ]
    }
   ],
   "source": [
    "history=model.fit(train_batches,validation_data=valid_batches,epochs=EPOCHS,callbacks=[redlr,chkpt,es],verbose=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 591
    },
    "id": "eTR__utZAdqe",
    "outputId": "f0d19fee-91c9-4fcb-b069-e148c8c1c4c7"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dict_keys(['loss', 'f1_metric', 'val_loss', 'val_f1_metric', 'lr'])\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "print(history.history.keys())\n",
    "\n",
    "plt.plot(history.history['f1_metric'])\n",
    "plt.plot(history.history['val_f1_metric'])\n",
    "plt.title('model accuracy')\n",
    "plt.ylabel('accuracy')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()\n",
    "\n",
    "plt.plot(history.history['loss'])\n",
    "plt.plot(history.history['val_loss'])\n",
    "plt.title('model loss')\n",
    "plt.ylabel('loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 35
    },
    "id": "SnW3bzknAdqi",
    "outputId": "088b94de-e4f7-4368-ac7e-248ad9b16349"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 500 images belonging to 2 classes.\n"
     ]
    }
   ],
   "source": [
    "test_batches=ImageDataGenerator(rescale=1/255).flow_from_directory('./T',target_size=IMG_SIZE,shuffle=False,class_mode=None,seed=SEED,batch_size=BATCHSIZE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "id": "IRV5160QMTf7"
   },
   "outputs": [],
   "source": [
    "# shutil.move(\"path/to/current/file\", \"path/to/new/destination/for/file\")\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "id": "H_cS6xCUMTiC"
   },
   "outputs": [],
   "source": [
    "# import os\n",
    "# import shutil\n",
    "# for _file in os.listdir('./data/DATA/TEST'):\n",
    "#   shutil.move(\"./data/DATA/TEST/\"+_file, \"./T/test/\"+_file)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "id": "xcbQFrWSMTkO"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "id": "WAtyFO2jMTm7"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {
    "id": "rKHkDIAzAdql"
   },
   "outputs": [],
   "source": [
    "model=load_model('best.h5',custom_objects={'f1_metric':f1_metric})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 35
    },
    "id": "LK6yZjZxAdqo",
    "outputId": "a3cfbc89-01b1-4881-a78f-c38ddebbf159"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "11/11 [==============================] - 5s 431ms/step\n"
     ]
    }
   ],
   "source": [
    "test_preds=model.predict_generator(test_batches,verbose=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "id": "hzzCVVU2Adqr"
   },
   "outputs": [],
   "source": [
    "test_preds=test_preds.argmax(axis=-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 563
    },
    "id": "ZDmv1bhfOboP",
    "outputId": "5567030a-29e6-4aa9-f9ac-7ac369ef5184"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([35, 43, 22, 32, 37, 43, 34, 29, 12,  4,  2, 34, 39, 38, 35, 37,  4,\n",
       "       30, 36, 18,  1, 40,  8, 25, 32, 37, 21, 28, 15, 34,  4, 29, 20, 10,\n",
       "       18,  6,  0,  4, 44, 40, 12, 44,  3, 41, 43, 36, 17, 35,  1, 42, 30,\n",
       "       22, 45,  4, 41,  7, 16, 42, 31, 25, 27,  1, 17, 17, 46,  6, 28,  5,\n",
       "        7, 13, 21, 48,  7, 22,  9, 21, 25, 49,  8,  0, 42, 10, 33, 37,  3,\n",
       "       14, 16, 34,  2, 42,  1, 28, 22, 12, 35, 37, 16, 33, 15, 47, 27, 21,\n",
       "        8,  3, 20, 37, 14, 40, 48,  9, 42, 31, 19, 14, 23, 27, 35,  1, 16,\n",
       "       47, 21,  9, 47, 35, 17, 15, 16, 36, 20, 26, 27, 20, 15, 49, 14, 37,\n",
       "       23,  6, 36, 44, 24, 48,  2,  6, 19, 43, 49, 27,  2, 39, 40, 26, 25,\n",
       "       31,  2,  8, 49, 42, 10, 28, 33, 44, 20, 47, 27, 28, 10, 38, 10,  9,\n",
       "       29, 16, 19, 23, 28, 46, 23, 32, 36, 24,  2,  7, 18, 23, 15, 11, 40,\n",
       "        9, 47, 37,  3, 12, 45, 23,  6, 33,  5, 18, 37, 11, 42, 30, 39, 41,\n",
       "       47, 31, 36,  2, 19, 39,  6, 24, 16, 35, 28,  0, 46, 43, 48,  5, 39,\n",
       "       44, 32, 38, 40, 49, 24, 28, 34, 20, 46, 16, 28, 25,  2, 41, 11, 45,\n",
       "       41, 40, 20,  7,  5,  8, 42, 26, 38, 10, 13,  0, 16, 48,  4, 34, 30,\n",
       "       40, 13, 16, 13, 32, 45, 35,  1, 12, 43, 23, 46, 25,  0,  3, 13, 45,\n",
       "       34, 11, 12, 10, 32, 10, 49, 23, 19, 21,  7, 17, 49, 28, 26, 21, 18,\n",
       "       40, 44, 18, 32, 48, 41, 34, 48,  9, 19, 27, 15,  3, 11,  7, 36, 42,\n",
       "       16,  9, 43, 48, 31, 45,  6, 45,  9, 34, 12, 44, 13, 24, 41, 31, 15,\n",
       "        0, 23, 44,  1, 13, 29, 18, 20, 43, 45,  0, 28, 29, 47, 46, 41, 36,\n",
       "       30, 19, 32,  0, 47,  8, 11, 30,  3,  1, 38,  6, 32,  3, 30, 29,  6,\n",
       "       37, 25, 27, 14, 24,  3, 31, 18, 16, 19, 35, 47, 39, 27, 22, 17, 22,\n",
       "       26,  7,  4, 18, 36, 15, 41,  4, 25, 10, 12, 37, 49, 45, 13, 30, 49,\n",
       "        8, 46, 15,  7, 39, 29, 39, 10, 24, 26, 12,  4, 44, 29,  1, 27, 22,\n",
       "        7, 40, 26, 29, 11, 25,  0, 42, 12, 37, 47, 30, 38, 14, 21, 26,  5,\n",
       "       43, 21,  4, 13, 46, 14, 17, 14,  6, 38,  5,  8, 20, 34, 27, 49, 22,\n",
       "       14, 23, 39, 33, 25, 46, 11, 38, 30, 24, 33, 11, 32, 22, 44,  4,  9,\n",
       "       26,  3, 31, 15, 35, 48,  5, 39,  5, 17, 38, 34, 45,  8, 20, 24, 46,\n",
       "        2, 41, 19, 42, 45, 14, 22, 24, 17,  0,  5, 33, 21,  8, 19, 31,  2,\n",
       "       18, 38,  1, 36, 29, 31, 43])"
      ]
     },
     "execution_count": 48,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_preds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {
    "id": "eYLTzNgvAdqu"
   },
   "outputs": [],
   "source": [
    "idx_to_cls = {v: k for k, v in train_batches.class_indices.items()}\n",
    "prediction_cls= np.vectorize(idx_to_cls.get)(test_preds)\n",
    "filenames_to_cls = list(zip(test_batches.filenames, prediction_cls))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {
    "id": "-UDwegAwAdqw"
   },
   "outputs": [],
   "source": [
    "data = pd.DataFrame(filenames_to_cls)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {
    "id": "HeZVvJQdAdq1"
   },
   "outputs": [],
   "source": [
    "data.columns = ['Filename', 'Labels']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 198
    },
    "id": "qJe5r9W3Adq8",
    "outputId": "f1af798b-7d6c-4176-bc33-e43290b3a83b"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Filename</th>\n",
       "      <th>Labels</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>test/000.jpg</td>\n",
       "      <td>red admiral</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>test/001.jpg</td>\n",
       "      <td>straited queen</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>test/002.jpg</td>\n",
       "      <td>metalmark</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>test/003.jpg</td>\n",
       "      <td>pipevine swallow</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>test/004.jpg</td>\n",
       "      <td>scarce swallow</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       Filename            Labels\n",
       "0  test/000.jpg       red admiral\n",
       "1  test/001.jpg    straited queen\n",
       "2  test/002.jpg         metalmark\n",
       "3  test/003.jpg  pipevine swallow\n",
       "4  test/004.jpg    scarce swallow"
      ]
     },
     "execution_count": 52,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "id": "eRqDuapjAdrC"
   },
   "outputs": [],
   "source": [
    "import os\n",
    "data['Filename'] = pd.DataFrame(os.listdir(\"./T/test/\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "id": "QoohAwWnAdrG"
   },
   "outputs": [],
   "source": [
    "file_names = os.listdir('./T/test/')\n",
    "file_names = sorted(file_names)\n",
    "data['Filename'] = file_names"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "id": "yQvYwcDiO-IH"
   },
   "outputs": [],
   "source": [
    "data['Filename'] = file_names"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 198
    },
    "id": "eHnbAtB1O_vC",
    "outputId": "f6591477-5537-4009-a2fa-75dc055fd4e5"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Filename</th>\n",
       "      <th>Labels</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>000.jpg</td>\n",
       "      <td>red admiral</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>001.jpg</td>\n",
       "      <td>straited queen</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>002.jpg</td>\n",
       "      <td>metalmark</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>003.jpg</td>\n",
       "      <td>pipevine swallow</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>004.jpg</td>\n",
       "      <td>scarce swallow</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  Filename            Labels\n",
       "0  000.jpg       red admiral\n",
       "1  001.jpg    straited queen\n",
       "2  002.jpg         metalmark\n",
       "3  003.jpg  pipevine swallow\n",
       "4  004.jpg    scarce swallow"
      ]
     },
     "execution_count": 56,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {
    "id": "lvlaYJuQAdrK"
   },
   "outputs": [],
   "source": [
    "data.to_csv('sub1_rev_with_aug_300px_drop5_v5.csv', index = False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "MYI8goShXh4O"
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "Butterfly Classification.ipynb",
   "provenance": []
  },
  "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": "3.8.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}