1_introduction_dataset_preprocessing.ipynb

{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Sieć służąca do klasyfikacji pastylek leków - wprowadzenie oraz przygotowanie datasetu\n",
    "Wykonał Bartłomiej Tarcholik\n",
    "\n",
    "## Opis projektu\n",
    "Moim planem było stworzenie sieci neuronowej służącej do klasyfikacji tabletek pod kątem ich kształtu. Moim planem jest stworzenie aplikacji dla osób z chorobami przewlekłymi, która przypominać będzie i kontrolować brane przez taką osobę leki. Ważną częścią tej aplikacji jest dobrze działająca sieć neuronowa, którą postarałem się tutaj stworzyć.\n",
    "\n",
    "### Dataset\n",
    "Zestaw danych składa się z 94 obrazków, które zostają automatycznie obrobione w celu stworzenia zestawu obrazków 64x64 piksele, na których uczy się sieć neuronowa. Niestety ilość obrazków jest bardzo ograniczona jako, że musiałem dataset tworzyć sam i z bardzo ograniczoną ilością czasu.\n",
    "\n",
    "### Historia projektu - co próbowałem?\n",
    "Rozpocząłem pracę od szukania właściwej technologii do zastosowania w projekcie. W tym celu przeszukałem możliwe istniejące rozwiązania detekcji i zainteresowały mnie przede wszystkim dwa - SSD oraz YOLOv7. Rozpocząłem pracę od SSD, bo wydawało się łatwiejsze, lecz niestety brak możliwych do modyfikacji parametrów zmusił mnie do porzucenia tej sieci na potrzeby projektu. Skierowałem się więc w stronę YOLO, którego Bag of Freebies oraz Bag of Specials dają dużo większe możliwości, lecz YOLO miało problemy same w sobie.\n",
    "\n",
    "Choćbym nieważne ile próbował, model nie tworzył inferencji na obrazach lub filmach, co koniec końców spowodowało moją decyzję o porzuceniu YOLO.\n",
    "\n",
    "### Koncept własnej detekcji\n",
    "Po porażce z wcześniej wspomnianymi metodami detekcji, musiałem zdecydować się na inną formę detekcji, której jednak nie udało mi się finalnie zaimplementować pod kątem wykrywania, lecz została wykorzystana do stworzenia datasetu. Po usłyszeniu ile czasu na analizę jednego obrazka zajmuje podejście Sliding Window, zdecydowałem się na wykorzystanie operacji morfologicznych oraz algorytmu wykrywania krawędzi dostarczanych przez pakiet OpenCV. Napisana została funkcja, która wykrywa obiekty po krawędziach i tworzy z nich kwadratowe bounding boxy, które mogą zostać użyte do wycięcia fragmentu analizowanego obrazka jeden po drugim. Skuteczność czasowa takiego rozwiązania pozwala przyspieszenie analizy jednego obrazka o dowolnej wielkości przynajmniej dwudziestokrotnie w porównaniu ze Sliding Window. Następnie każdy wycinek może zostać przeskalowany i podany do klasyfikacji.\n",
    "Klasyfikacja daje możliwość sprawdzenia pewności dzięki funkcji wyjściowej, tak więc konkretny wycinek będzie uznawany za brak tabletki w przypadku, w którym pewność nie będzie wystarczająca.\n",
    "\n",
    "## UWAGA: Błąd został popełniony\n",
    "Jak piszę pod koniec projektu, popełniłem pewien poważny błąd. Początkowo obrazy przeskalowałem do wielkości 75x75 pikseli, jednak sieć trenuję na 64x64 pikselach. Ten błąd jest prawdopodobnie dosyć znaczący. Dla potrzeb uruchomienia istniejącego kodu, poprawiłem go tutaj, lecz wszystkie wyniki są dla obrazów 75x75 pikseli używanych przez sieć, która skaluje je do 64x64 pikseli, co niewątpliwie powoduje poważną utratę danych obrazów."
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Algorytm Bounding Boxów\n",
    "Funkcja GetBoundingBoxes jest funkcją gotową do zaimplementowania w dowolnym miejscu, zostanie wykorzystana na pewno w dalszym rozwoju projektu. Wykonuje ona następujące kroki:\n",
    "1. Pobranie obrazu i przeskalowanie go do mniejszego rozmiaru, żeby algorytm łatwiej wykrywał krawędzie\n",
    "2. Nałożenie rozmycia Gaussowskiego\n",
    "3. Zastosowanie algorytmu Canny'ego w celu znalezienia krawędzi na obrazie\n",
    "4. Dylatacja w celu pogrubienia krawędzi i połączenia dziur w nich\n",
    "5. Znalezienie konturów\n",
    "6. Wyliczenie Bounding Boxów dla każdego konturu\n",
    "7. Usunięcie nadmiarowych Bounding Boxów"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Edges:\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dilatation:\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import cv2\n",
    "import numpy as np\n",
    "from matplotlib import pyplot as plt\n",
    "import os\n",
    "\n",
    "def Save(directory, fileName, img):\n",
    "    if not os.path.exists(directory):\n",
    "        os.makedirs(directory)\n",
    "    cv2.imwrite(directory + \"/\" + fileName, img)\n",
    "\n",
    "def GetBoundingBoxes(image, sizeToScale, t1, t2, testMode=False):\n",
    "    x, y, z = image.shape\n",
    "    newX = 0\n",
    "    newY = 0\n",
    "    if x > y:\n",
    "        newX = sizeToScale\n",
    "        scale = newX / x\n",
    "        newY = round(y * scale)\n",
    "    else:\n",
    "        newY = sizeToScale\n",
    "        scale = newY / y\n",
    "        newX = round(x * scale)\n",
    "    res = cv2.resize(image, (newY, newX), interpolation = cv2.INTER_AREA)\n",
    "    img_blur = cv2.GaussianBlur(res, (3,3), 0, 0)\n",
    "    edges = cv2.Canny(img_blur, threshold1=t1, threshold2=t2)\n",
    "    if testMode:\n",
    "        print(\"Edges:\")\n",
    "        plt.imshow(edges)\n",
    "        plt.show()\n",
    "    im_X, im_Y = edges.shape\n",
    "    kernel = np.ones((5,5),np.uint8)\n",
    "    dilation = cv2.dilate(edges,kernel,iterations = 2)\n",
    "    if testMode:\n",
    "        print(\"Dilatation:\")\n",
    "        plt.imshow(dilation)\n",
    "        plt.show()\n",
    "    ret, thresh = cv2.threshold(dilation, 127, 255, 0)\n",
    "    contours, _ = cv2.findContours(thresh, cv2.RETR_TREE,\n",
    "                                cv2.CHAIN_APPROX_SIMPLE)\n",
    "    bboxes = []\n",
    "    for i in range(len(contours)):\n",
    "        cnt = contours[i]\n",
    "        x, y, w, h = cv2.boundingRect(cnt)\n",
    "        if w / im_X > 0.1 and h / im_Y:\n",
    "            bbox = [[round(x / scale), round(y / scale)], [round((x + w)  / scale), round((y + h)  / scale)]]\n",
    "            bboxes.append(bbox)\n",
    "    toMerge = []\n",
    "    unchanged = False\n",
    "    while unchanged == False:\n",
    "        unchanged = True\n",
    "        for i in range(len(bboxes)):\n",
    "            if unchanged:\n",
    "                bbox = bboxes[i]\n",
    "                start = bbox[0]\n",
    "                end = bbox[1]\n",
    "                add = True\n",
    "                for j in range(len(bboxes)):\n",
    "                    if i != j and unchanged:\n",
    "                        other = bboxes[j]\n",
    "                        st = other[0]\n",
    "                        en = other[1]\n",
    "                        dx = min(end[0], en[0]) - max(start[0], st[0])\n",
    "                        dy = min(end[1], en[1]) - max(start[1], st[1])\n",
    "                        if (dx>=0) and (dy>=0):\n",
    "                            inter_area = dx * dy\n",
    "                            smaller = min(\n",
    "                            (end[0] - start[0]) * (end[1] - start[1]),\n",
    "                            (en[0] - st[0]) * (en[1] - st[1])\n",
    "                            )\n",
    "                            if inter_area > (0.5 * smaller):\n",
    "                                bboxes[i][0] = [min(start[0], st[0]), min(start[1], st[1])]\n",
    "                                bboxes[i][1] = [max(end[0], en[0]), max(end[1], en[1])]\n",
    "                                bboxes.pop(j)\n",
    "                                unchanged = False\n",
    "    for bbox in bboxes:\n",
    "        xdif = bbox[1][0] - bbox[0][0]\n",
    "        ydif = bbox[1][1] - bbox[0][1]\n",
    "        val = round(abs(xdif - ydif) / 2)\n",
    "        if xdif > ydif:\n",
    "            y_start = bbox[0][1] - val\n",
    "            y_end = bbox[1][1] + val\n",
    "            if y_start < 0:\n",
    "                offset = abs(y_start)\n",
    "                y_start = 0\n",
    "                y_end = y_end + offset\n",
    "            elif y_end > image.shape[1]:\n",
    "                offset = y_end - image.shape[1]\n",
    "                y_end = image.shape[1]\n",
    "                y_start = y_start - offset\n",
    "            bbox[0][1] = y_start\n",
    "            bbox[1][1] = y_end\n",
    "        else:\n",
    "            x_start = bbox[0][0] - val\n",
    "            x_end = bbox[1][0] + val\n",
    "            if x_start < 0:\n",
    "                offset = abs(x_start)\n",
    "                x_start = 0\n",
    "                x_end = x_end + offset\n",
    "            elif x_end > image.shape[0]:\n",
    "                offset = x_end - image.shape[0]\n",
    "                x_end = image.shape[0]\n",
    "                x_start = x_start - offset\n",
    "            bbox[0][0] = x_start\n",
    "            bbox[1][0] = x_end\n",
    "    return bboxes\n",
    "\n",
    "img = cv2.imread(\"C:/ml/data/pills/original/0028_square_white.jpg\")\n",
    "img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)\n",
    "plt.imshow(img)\n",
    "plt.show()\n",
    "\n",
    "bx = GetBoundingBoxes(img, 100, 10, 100, True)[0]\n",
    "plt.imshow(img[bx[0][1]:bx[1][1], bx[0][0]:bx[1][0]])\n",
    "plt.show()"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Edycja datasetu\n",
    "Przy użyciu powyższej funkcji każde zdjęcie jest obcinane i przerabiane na nowe. Początkowo były one w 2 różnych miejscach z różnymi konwencjami nazewniczymi, stąd konieczność podzielenia ich na dwie pętle.\n",
    "Z każdego zdjęcia wycinana jest tabletka, wycinek jest zapisywany jako obraz 64x64 piksele."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dataset has been updated, total changes made: 94\n"
     ]
    }
   ],
   "source": [
    "dataset_dir = 'C:/ml/data/pills/original'\n",
    "uploaded_dir = 'C:/ml/data/pills/uploaded'\n",
    "base_dir = 'C:/ml/data/pills/preprocessed'\n",
    "full_dir = 'C:/ml/data/pills/preprocessed/full'\n",
    "final_dir = 'C:/ml/data/pills/final'\n",
    "if not os.path.exists(base_dir):\n",
    "    os.mkdir(base_dir)\n",
    "\n",
    "if not os.path.exists(full_dir):\n",
    "    os.mkdir(full_dir)\n",
    "    \n",
    "og_names = [f for f in os.listdir(dataset_dir) if f.endswith('.jpg')]\n",
    "\n",
    "changes = 0\n",
    "for fname in og_names:\n",
    "    nameData = fname.split(\"_\")\n",
    "    nameData[-1] = nameData[-1][0:-4]\n",
    "    image = cv2.imread(dataset_dir + \"/\" + fname)\n",
    "    image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)\n",
    "    bboxes = GetBoundingBoxes(image, 100, 50, 200)\n",
    "    if len(bboxes) == 0:\n",
    "            gray = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)\n",
    "            bboxes = GetBoundingBoxes(image, 100, 10, 100)\n",
    "            if len(bboxes) != 1:\n",
    "                print(fname, len(bboxes))\n",
    "                plt.imshow(image)\n",
    "                plt.show()\n",
    "            else:\n",
    "                bx = bboxes[0]\n",
    "                img = image[bx[0][1]:bx[1][1], bx[0][0]:bx[1][0]]\n",
    "                res = cv2.resize(img, (64, 64), interpolation = cv2.INTER_AREA)\n",
    "                Save(final_dir + \"/\" + nameData[1], str(changes) + \".jpg\", res)\n",
    "                changes += 1\n",
    "    else:\n",
    "        bx = bboxes[0]\n",
    "        img = image[bx[0][1]:bx[1][1], bx[0][0]:bx[1][0]]\n",
    "        res = cv2.resize(img, (64, 64), interpolation = cv2.INTER_AREA)\n",
    "        Save(final_dir + \"/\" + nameData[1], str(changes) + \".jpg\", res)\n",
    "        changes += 1\n",
    "\n",
    "up_dirs = [f for f in os.listdir(uploaded_dir)]\n",
    "\n",
    "for dir in up_dirs:\n",
    "    fnames = [f for f in os.listdir(uploaded_dir + \"/\" + dir)]\n",
    "    for fname in fnames:\n",
    "        image = cv2.imread(uploaded_dir + \"/\" + dir + \"/\" + fname)\n",
    "        image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)\n",
    "        bboxes = GetBoundingBoxes(image, 100, 50, 200)\n",
    "        if len(bboxes) == 0:\n",
    "            gray = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)\n",
    "            bboxes = GetBoundingBoxes(image, 100, 10, 100)\n",
    "            if len(bboxes) != 1:\n",
    "                print(dir + \"/\" + fname, len(bboxes))\n",
    "                plt.imshow(image)\n",
    "                plt.show()\n",
    "            else:\n",
    "                bx = bboxes[0]\n",
    "                img = image[bx[0][1]:bx[1][1], bx[0][0]:bx[1][0]]\n",
    "                res = cv2.resize(img, (64, 64), interpolation = cv2.INTER_AREA)\n",
    "                Save(final_dir + \"/\" + dir, str(changes) + \".jpg\", res)\n",
    "                changes += 1\n",
    "        else:\n",
    "            bx = bboxes[0]\n",
    "            img = image[bx[0][1]:bx[1][1], bx[0][0]:bx[1][0]]\n",
    "            res = cv2.resize(img, (64, 64), interpolation = cv2.INTER_AREA)\n",
    "            Save(final_dir + \"/\" + dir, str(changes) + \".jpg\", res)\n",
    "            changes += 1\n",
    "\n",
    "print(\"Dataset has been updated, total changes made:\", changes)\n"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Nadrabianie małej ilości zdjęć\n",
    "W celu nadrobienia małej ilości zdjęć, używam ImageDataGeneratora by stworzyć dużą ich ilość w celach treningowych. Każde zdjęcie jest zniekształcane i zapisywane wiele razy. Dzięki temu każdy kształt posiada przynajmniej 1000 różnych zdjęć do treningu oraz osobno modyfikowane zdjęcia do walidacji. Niestety oznacza to jednak, że jest spore prawdopodobieństwo fałszywej klasyfikacji po podaniu zdjęcia np. kwadratowej tabletki, gdyż niektóre kształty operowały na ekstremalnie małej ilości zdjęć początkowych."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "C:/ml/data/pills/final/capsule\n",
      "C:/ml/data/pills/final/flat-capsule\n",
      "C:/ml/data/pills/final/oval\n",
      "C:/ml/data/pills/final/oval-capsule\n",
      "C:/ml/data/pills/final/round\n",
      "C:/ml/data/pills/final/square\n"
     ]
    }
   ],
   "source": [
    "from tensorflow.keras.preprocessing import image\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "img_rows = 64\n",
    "img_cols = 64\n",
    "\n",
    "datagen = ImageDataGenerator(\n",
    "    rotation_range=90,\n",
    "    width_shift_range=0.2,\n",
    "    height_shift_range=0.2,\n",
    "    shear_range=0.2,\n",
    "    zoom_range=0.2,\n",
    "    channel_shift_range=50,\n",
    "    brightness_range=[0.4, 1.6],\n",
    "    horizontal_flip=True,\n",
    "    fill_mode='nearest')\n",
    "\n",
    "train_dir = 'C:/ml/data/pills/deploy/train'\n",
    "val_dir = 'C:/ml/data/pills/deploy/val'\n",
    "dirs = [f for f in os.listdir(final_dir)]\n",
    "v = 0\n",
    "for directory in dirs:\n",
    "    cur_dir = final_dir + \"/\" + directory\n",
    "    print(cur_dir)\n",
    "    images = [f for f in os.listdir(cur_dir) if f.endswith('.jpg')]\n",
    "    toClone = round(1000 / len(images)) + 1\n",
    "    for img_path in images:\n",
    "        im = cv2.imread(cur_dir + \"/\" + img_path)\n",
    "        im = cv2.cvtColor(im, cv2.COLOR_BGR2RGB)\n",
    "        Save(train_dir + \"/\" + directory, str(changes) + \".jpg\", im)\n",
    "        changes += 1\n",
    "        img = image.load_img(cur_dir + \"/\" + img_path, target_size=(img_rows, img_cols))\n",
    "        x = image.img_to_array(img)\n",
    "        x = x.reshape((1,) + x.shape)\n",
    "        i = 0\n",
    "        for batch in datagen.flow(x, batch_size=1):\n",
    "            new_image = np.array(image.array_to_img(batch[0]))\n",
    "            Save(train_dir + \"/\" + directory, str(changes) + \".jpg\", new_image)\n",
    "            i += 1\n",
    "            changes += 1\n",
    "            if i % toClone == 0:\n",
    "                break\n",
    "        i = 0\n",
    "        for batch in datagen.flow(x, batch_size=1):\n",
    "            new_image = np.array(image.array_to_img(batch[0]))\n",
    "            Save(val_dir + \"/\" + directory, str(v) + \".jpg\", new_image)\n",
    "            i += 1\n",
    "            v += 1\n",
    "            if i % toClone == 0:\n",
    "                break"
   ]
  },
  {
   "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": "3.10.7 (tags/v3.10.7:6cc6b13, Sep  5 2022, 14:08:36) [MSC v.1933 64 bit (AMD64)]"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "1db53b7b098ce13369892f6fa64e6960b1b60742cc9b3126941b620089625937"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}