AI_hackathon2.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "AI hackathon.ipynb",
      "provenance": [],
      "mount_file_id": "1kPrE_gCSVx2yofva-X4bgP2u_sKTfYpf",
      "authorship_tag": "ABX9TyNCer/GibL4Jg5kBz/5B4r7",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/GmBeHappy/AIAT-AI-HACKATHON-COVID-19-recognize-x-ray/blob/main/AI_hackathon2.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SN4T1Qbuf961",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 36
        },
        "outputId": "e41015af-ed08-4d24-8ac8-793c8f9775e3"
      },
      "source": [
        "import cv2\n",
        "#อิมพอร์ต cv2 ที่มาจาก OpenCV\n",
        "import matplotlib.pyplot as plt\n",
        "#อิมพอร์ต pyplot เข้ามาในชื่อ plt\n",
        "from imutils import paths\n",
        "from keras import backend\n",
        "from keras.models import Sequential\n",
        "from keras.layers.convolutional import Conv2D\n",
        "from keras.layers.core import Dense, Flatten\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.metrics import accuracy_score, classification_report\n",
        "import csv"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "cmDcUFJMgicL",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 288
        },
        "outputId": "e66e4dfb-0f79-496a-cf5c-65b95f7681f1"
      },
      "source": [
        "image_paths = list(paths.list_images(\"drive/My Drive/dataset ai hackathon\"))\n",
        "#อ่านรูปาภาพสัตว์ทั้งหมดจากโฟลเดอร์\n",
        "image = cv2.imread(image_paths[10])\n",
        "#ลองใช้ cv2 อ่านรูปขึ้นมาสักรูป\n",
        "new_image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)\n",
        "#หลังจากอ่านมาเสร็จแล้ว เราต้องแปลงสีก่อนเพราะ cv2 นั้นเมื่ออ่านรูปเข้ามา จะมีค่าพื้นฐานของสีเป็น BGR ซึ่งสีที่เราใช้ๆกันอยู่อย่างคุ้นตาคือ RGB เพราะฉะนั้นเราต้องแปลงก่อน\n",
        "plt.figure()\n",
        "#เราใช้ pyplot ในการโชว์รูปขึ้นมา แต่ก่อนที่เราจะโชว์ขึ้นมาได้เราต้องสร้างตัว figure ขึ้นมาเสียก่อน เหมือนกับการสร้างหน้าต่างโปรแกรมขึ้นมาเพื่อที่จะให้เอารูปที่จะโชว์ไปแปะ\n",
        "plt.imshow(new_image)\n",
        "#ใช้ฟังก์ชั่น imshow ในการโชว์รูปออกมา"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7faa5acb68d0>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 2
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hxQcSlP-gpwH",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 746
        },
        "outputId": "41154087-de9a-40ad-e1bb-b62e760a8f86"
      },
      "source": [
        "import random\n",
        "#อิมพอร์ต random เข้ามาเพื่อที่เราจะใช้ฟังก์ชั่น shuffle\n",
        "import numpy as np\n",
        "#อิมพอร์ต numpy เข้ามาในชื่อ np\n",
        "random.shuffle(image_paths)\n",
        "#ใช้ฟังก์ชั่น shuffle ในการสลับรูปภาพเพื่อให้ภาพมันไม่เรียงติดกันเหมือนเดิมในทุกครั้ง\n",
        "data = []\n",
        "labels = []\n",
        "#สร้างตัวแปรอาเรย์ไว้รอเก็บค่าจากการแปลงรูปเป็นข้อมูล โดย data จะใช้เก็บข้อมูลเทรน ส่วน labels จะใช้เก็บเป็นข้อมูลสอน/คำตอบ\n",
        "for image_path in image_paths:\n",
        "#วนลูปเพื่อทำการแปลงค่าจากรูปภาพเป็นข้อมูล\n",
        "    image = cv2.imread(image_path)\n",
        "    #อ่านรูปจาก image_path\n",
        "    image = cv2.resize(image, (32, 32))\n",
        "    #ปรับไซส์ให้รูปภาพมีคุณภาพต่ำลง(ตอนเทรนนิ่งข้อมูล คอมเราจะได้ไม่ทำงานหนักเกินไป)\n",
        "    labels.append(image_path.split(\"/\")[-2])\n",
        "    #เอาชื่อ image_path ของแต่ละรูปมาเก็บค่าลงตัวแปรอาเรย์ labels\n",
        "    data.append(image)\n",
        "    #เอาข้อมูลรูปภาพใส่ตัวแปรอาเรย์ data เก็บไว้\n",
        "data = np.array(data)\n",
        "labels = np.array(labels)\n",
        "from sklearn.preprocessing import LabelBinarizer\n",
        "lb = LabelBinarizer()\n",
        "#อิมพอร์ต LabelBinarizer ของ sklearn มาใช้\n",
        "labels = lb.fit_transform(labels)\n",
        "#เรียกใช้ฟังก์ชั่น fit_transform เพื่อแปลง labels เป็นเลขฐานสอง\n",
        "\n",
        "X_train, X_test, y_train, y_test = train_test_split(data, labels, test_size=0.1, random_state=5)\n",
        "#เรียกใช้ฟังก์ชั่น train_test_split เพื่อทำการสร้างข้อมูลเทรนกับเทส\n",
        "model = Sequential()\n",
        "#ใช้โมเดล Sequential ของ keras\n",
        "model.add(Conv2D(32, (3, 3), input_shape=(32, 32 ,3), activation='relu', padding='same'))\n",
        "model.add(Conv2D(64, (3, 3), activation='relu', padding='same'))\n",
        "#ใข้คำสั่ง model.add() เพิ่มชั้น(Layer)ให้กับโมเดลโดยให้ชั้นนี้เป็น convolution layer ด้วยฟังก์ชั่น Conv2D โดยค่าพารามิเตอร์แรกคือ filters มีค่าเป็น 32 คือบอกว่า คือจำนวนฟิลเตอร์ทั้งหมดของเลเยอร์นี้มี 32 ชั้น, พารามิเตอร์ตัวที่สองคือ kernel_size มีค่าเป็น 3x3 และ input_shape=(32, 32 ,3) คือรูปขนาด 32x32 และมี 3 channels ในนี้ก็คือ RGB, activation function เป็น relu, และ padding เซ็ทไว้เป็น same  เพื่อให้ข้อมูลที่ส่งออกมีขนาดเท่าเดิม (ไม่ให้ output ที่ได้มีขนาดเล็กลง)\n",
        "model.add(Flatten())\n",
        "#เพิ่มชั้น(Layer) ที่เป็น flatten ให้กับโมเดลเพื่อแปลงข้อมูลจากภาพให้เป็นเวคเตอร์ ที่จะสามารถส่งต่อให้ชั้นต่อไปได้ (ง่ายๆคือเอาข้อมูลอาร์เรย์ขนาดหลายมิติจากชั้นที่แล้ว มาตีแผ่ออกเหลือแค่มิติเดียว)\n",
        "model.add(Dense(3, activation='softmax'))\n",
        "#เพิ่มชั้น(Layer) ที่เป็น dense ให้กับโมเดลซึ่งก็คือชั้นที่เป็น fully connected นั่นเอง และใช้ activation function เป็น softmax และมี output เป็นอาร์เรย์ขนาด (*, 3)\n",
        "model.summary()\n",
        "model.compile(loss='categorical_crossentropy', optimizer='adam')\n",
        "model.fit(X_train, y_train, epochs=10, validation_split=0.2)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_1\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d_1 (Conv2D)            (None, 32, 32, 32)        896       \n",
            "_________________________________________________________________\n",
            "conv2d_2 (Conv2D)            (None, 32, 32, 64)        18496     \n",
            "_________________________________________________________________\n",
            "flatten_1 (Flatten)          (None, 65536)             0         \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 3)                 196611    \n",
            "=================================================================\n",
            "Total params: 216,003\n",
            "Trainable params: 216,003\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "Train on 550 samples, validate on 138 samples\n",
            "Epoch 1/10\n",
            "550/550 [==============================] - 3s 5ms/step - loss: 84.6524 - val_loss: 1.7709\n",
            "Epoch 2/10\n",
            "550/550 [==============================] - 2s 4ms/step - loss: 1.0658 - val_loss: 1.2692\n",
            "Epoch 3/10\n",
            "550/550 [==============================] - 2s 4ms/step - loss: 0.7446 - val_loss: 1.2640\n",
            "Epoch 4/10\n",
            "550/550 [==============================] - 2s 4ms/step - loss: 0.5586 - val_loss: 1.2863\n",
            "Epoch 5/10\n",
            "550/550 [==============================] - 2s 4ms/step - loss: 0.4345 - val_loss: 1.2781\n",
            "Epoch 6/10\n",
            "550/550 [==============================] - 2s 4ms/step - loss: 0.2992 - val_loss: 1.7765\n",
            "Epoch 7/10\n",
            "550/550 [==============================] - 2s 4ms/step - loss: 0.2325 - val_loss: 1.5393\n",
            "Epoch 8/10\n",
            "550/550 [==============================] - 2s 4ms/step - loss: 0.1702 - val_loss: 1.7971\n",
            "Epoch 9/10\n",
            "550/550 [==============================] - 2s 4ms/step - loss: 0.1565 - val_loss: 1.7174\n",
            "Epoch 10/10\n",
            "550/550 [==============================] - 2s 4ms/step - loss: 0.1161 - val_loss: 2.0562\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.callbacks.History at 0x7faa5a7baeb8>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Sarzfi0Bg3nt",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 209
        },
        "outputId": "50fcd214-807c-46a8-e09b-1f239466f8ff"
      },
      "source": [
        "#print(X_test)\n",
        "y_pred = model.predict(X_test)\n",
        "#หลังจากที่เราเทรนตัวโมเดลเราเสร็จ คราวนี้เราจะใช้ข้อมูลเทสมาลองให้โมเดลมัน classified โดยใช้ฟังก์ชั่น predict\n",
        "from sklearn.metrics import accuracy_score, classification_report #อิมพอร์ต accuracy_score กับ classification_report จาก sklearn.metrics \n",
        "accuracy_score(y_test.argmax(axis=1), y_pred.argmax(axis=1))\n",
        "#จากนั้นมาเรียกใช้ฟังก์ชั่น accuracy_score เพื่อดูความแม่นของข้อมูลที่ predict กัน\n",
        "print(classification_report(y_test.argmax(axis=1), y_pred.argmax(axis=1), target_names=lb.classes_))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "       covid       0.59      0.43      0.50        23\n",
            "      normal       0.45      0.76      0.57        25\n",
            "   pneumonia       0.44      0.28      0.34        29\n",
            "\n",
            "    accuracy                           0.48        77\n",
            "   macro avg       0.50      0.49      0.47        77\n",
            "weighted avg       0.49      0.48      0.46        77\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6Ja0f34bhA3-",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "1cc7de23-76f5-4b66-90ab-2dc8047aacc3"
      },
      "source": [
        "test_paths = list(paths.list_images(\"drive/My Drive/test_release\"))\n",
        "data2 = []\n",
        "#สร้างตัวแปรอาเรย์ไว้รอเก็บค่าจากการแปลงรูปเป็นข้อมูล โดย data จะใช้เก็บข้อมูลเทรน ส่วน labels จะใช้เก็บเป็นข้อมูลสอน/คำตอบ\n",
        "for test_path in test_paths:\n",
        "#วนลูปเพื่อทำการแปลงค่าจากรูปภาพเป็นข้อมูล\n",
        "    image = cv2.imread(test_path)\n",
        "    #อ่านรูปจาก image_path\n",
        "    image = cv2.resize(image, (32, 32))\n",
        "    #ปรับไซส์ให้รูปภาพมีคุณภาพต่ำลง(ตอนเทรนนิ่งข้อมูล คอมเราจะได้ไม่ทำงานหนักเกินไป)\n",
        "\n",
        "    data2.append(image)\n",
        "    #เอาข้อมูลรูปภาพใส่ตัวแปรอาเรย์ data เก็บไว้\n",
        "datatest = np.array(data2)\n",
        "\n",
        "\n",
        "y_pred = model.predict_classes(datatest)\n",
        "num=1\n",
        "\n",
        "def sendout():\n",
        "  row_list = [[\"Id\", \"Predicted\"]]\n",
        "  for i in range(len(datatest)):\n",
        "    print(\"%s=%s\" % (i+1,y_pred[i]))\n",
        "    if(i+1<10):\n",
        "      dt=\"00\"+str(i+1)\n",
        "    elif(i+1<100):\n",
        "      dt=\"0\"+str(i+1)\n",
        "    else:\n",
        "      dt=str(i+1)\n",
        "    row_list.append([\"test\"+dt,str(y_pred[i])])\n",
        "  return row_list\n",
        "sendout()\n",
        "#print(sendout())\n",
        "\n",
        "csv.register_dialect('myDialect',\n",
        "                     delimiter=',',\n",
        "                     quoting=csv.QUOTE_ALL)\n",
        "with open('finallyyeah.csv', 'w', newline='') as file:\n",
        "    writer = csv.writer(file, dialect='myDialect')\n",
        "    writer.writerows(sendout())"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "1=2\n",
            "2=1\n",
            "3=1\n",
            "4=2\n",
            "5=0\n",
            "6=1\n",
            "7=1\n",
            "8=1\n",
            "9=1\n",
            "10=1\n",
            "11=1\n",
            "12=1\n",
            "13=1\n",
            "14=1\n",
            "15=1\n",
            "16=0\n",
            "17=1\n",
            "18=1\n",
            "19=1\n",
            "20=2\n",
            "21=2\n",
            "22=2\n",
            "23=2\n",
            "24=1\n",
            "25=0\n",
            "26=1\n",
            "27=1\n",
            "28=0\n",
            "29=1\n",
            "30=1\n",
            "31=0\n",
            "32=1\n",
            "33=0\n",
            "34=1\n",
            "35=1\n",
            "36=1\n",
            "37=1\n",
            "38=2\n",
            "39=1\n",
            "40=0\n",
            "41=1\n",
            "42=1\n",
            "43=1\n",
            "44=0\n",
            "45=2\n",
            "46=1\n",
            "47=2\n",
            "48=1\n",
            "49=0\n",
            "50=1\n",
            "51=1\n",
            "52=2\n",
            "53=1\n",
            "54=1\n",
            "55=2\n",
            "56=2\n",
            "57=0\n",
            "58=0\n",
            "59=1\n",
            "60=2\n",
            "61=0\n",
            "62=2\n",
            "63=0\n",
            "64=1\n",
            "65=1\n",
            "66=1\n",
            "67=1\n",
            "68=2\n",
            "69=2\n",
            "70=0\n",
            "71=1\n",
            "72=0\n",
            "73=2\n",
            "74=1\n",
            "75=0\n",
            "76=2\n",
            "77=2\n",
            "78=1\n",
            "79=1\n",
            "80=1\n",
            "81=0\n",
            "82=0\n",
            "83=0\n",
            "84=0\n",
            "85=1\n",
            "1=2\n",
            "2=1\n",
            "3=1\n",
            "4=2\n",
            "5=0\n",
            "6=1\n",
            "7=1\n",
            "8=1\n",
            "9=1\n",
            "10=1\n",
            "11=1\n",
            "12=1\n",
            "13=1\n",
            "14=1\n",
            "15=1\n",
            "16=0\n",
            "17=1\n",
            "18=1\n",
            "19=1\n",
            "20=2\n",
            "21=2\n",
            "22=2\n",
            "23=2\n",
            "24=1\n",
            "25=0\n",
            "26=1\n",
            "27=1\n",
            "28=0\n",
            "29=1\n",
            "30=1\n",
            "31=0\n",
            "32=1\n",
            "33=0\n",
            "34=1\n",
            "35=1\n",
            "36=1\n",
            "37=1\n",
            "38=2\n",
            "39=1\n",
            "40=0\n",
            "41=1\n",
            "42=1\n",
            "43=1\n",
            "44=0\n",
            "45=2\n",
            "46=1\n",
            "47=2\n",
            "48=1\n",
            "49=0\n",
            "50=1\n",
            "51=1\n",
            "52=2\n",
            "53=1\n",
            "54=1\n",
            "55=2\n",
            "56=2\n",
            "57=0\n",
            "58=0\n",
            "59=1\n",
            "60=2\n",
            "61=0\n",
            "62=2\n",
            "63=0\n",
            "64=1\n",
            "65=1\n",
            "66=1\n",
            "67=1\n",
            "68=2\n",
            "69=2\n",
            "70=0\n",
            "71=1\n",
            "72=0\n",
            "73=2\n",
            "74=1\n",
            "75=0\n",
            "76=2\n",
            "77=2\n",
            "78=1\n",
            "79=1\n",
            "80=1\n",
            "81=0\n",
            "82=0\n",
            "83=0\n",
            "84=0\n",
            "85=1\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}