{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"name":"python","version":"3.7.12","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"code","source":"# This Python 3 environment comes with many helpful analytics libraries installed\n# It is defined by the kaggle/python Docker image: https://github.com/kaggle/docker-python\n# For example, here's several helpful packages to load\n\nimport numpy as np # linear algebra\nimport pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n\n# Input data files are available in the read-only \"../input/\" directory\n# For example, running this (by clicking run or pressing Shift+Enter) will list all files under the input directory\n\nimport os\nfor dirname, _, filenames in os.walk('/kaggle/input'):\n    for filename in filenames:\n        print(os.path.join(dirname, filename))\n\n# You can write up to 20GB to the current directory (/kaggle/working/) that gets preserved as output when you create a version using \"Save & Run All\" \n# You can also write temporary files to /kaggle/temp/, but they won't be saved outside of the current session","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# multiclass classification for isup grade\n\nimport numpy as np\nimport os\nfrom PIL import Image\nfrom matplotlib import pyplot as plt\nimport pandas as pd\nfrom imblearn.under_sampling import RandomUnderSampler\nrus = RandomUnderSampler(random_state=0)\n\ndef get_data(data_dir):\n    data = [] \n    path = os.path.join(data_dir)\n    for img in os.listdir(path):\n        if img.endswith('.png'):\n            img = Image.open(os.path.join(path, img)).convert(\"L\")\n            img.load()\n            img.thumbnail((64,64), Image.ANTIALIAS)\n            array = np.asarray(img, dtype=\"int32\")\n            data.append(array)                       \n    return np.array(list(data))\n\nX = get_data(\"/kaggle/input/panda-resized-train-data-512x512/train_images/train_images\")\n# X_test = get_data('/kaggle/input/panda-resized-tt/test_images/test_images')\n\nprint(\"gata x\")\ndf = pd.read_csv(\"/kaggle/input/prostate-cancer-grade-assessment/train.csv\")\nY = df['isup_grade'].astype(str).to_numpy()\n\nX_sm,Y_sm = rus.fit_resample(X.reshape(X.shape[0], -1) ,Y)\nX_sm.reshape(-1, 64, 64, 1)\n# df1 = pd.read_csv(\"/kaggle/input/panda-resized-tt/test.csv\")\n# Y_test = df1['isup_grade'].astype(str).to_numpy()\n\nfrom tensorflow.keras.utils import to_categorical\n\nX = np.expand_dims(X, axis=3)\nY = to_categorical(Y)\n# X_test = np.expand_dims(X_test, axis=3)\n# Y_test = to_categorical(Y_test)\n\n\nfrom tensorflow import keras\nfrom tensorflow.keras.models import Sequential\nfrom tensorflow.keras.layers import Dense, Dropout, Flatten\nfrom tensorflow.keras.layers import Conv2D, MaxPool2D\n\nnum_output_classes = 6\ninput_img_size = (64, 64, 1)  # 64x64 image with 1 color channel\n\nmodel= Sequential()\nmodel.add(Conv2D(kernel_size=(3,3), filters=32, activation='tanh', input_shape=input_img_size))\nmodel.add(Conv2D(filters=30,kernel_size = (3,3),activation='tanh'))\nmodel.add(MaxPool2D(2,2))\nmodel.add(Conv2D(filters=30,kernel_size = (3,3),activation='tanh'))\nmodel.add(MaxPool2D(2,2))\nmodel.add(Conv2D(filters=30,kernel_size = (3,3),activation='tanh'))\n\nmodel.add(Flatten())\n\nmodel.add(Dense(20,activation='relu'))\nmodel.add(Dense(15,activation='relu'))\nmodel.add(Dense(6,activation = 'softmax'))\n\nmodel.compile(\n    loss=keras.losses.categorical_crossentropy,\n    optimizer=keras.optimizers.Adam(),\n    metrics=[\"accuracy\"],\n)\n\nbatch_size = 128\nepochs = 2\n\nhistory  = model.fit(X_sm, Y_sm,\n          batch_size=50,\n          epochs=epochs,\n          verbose=1,\n          validation_split=0.2)\n\nimg = Image.open('/kaggle/input/panda-resized-tt/test_images/test_images/0018ae58b01bdadc8e347995b69f99aa.png').convert(\"L\")\nimg.load()\nimg.thumbnail((64,64), Image.ANTIALIAS)\nmi = np.asarray(img, dtype=\"int32\")\nmi = np.expand_dims(mi, axis=0)\nmi = np.expand_dims(mi, axis=3)\nprediction = model.predict(mi)\nprint([x for x in prediction])\n\n# score = model.evaluate(X_test, Y_test, verbose=0)\n# print('Test loss:', score[0])\n# print('Test accuracy:', score[1])\nmodel_path ='/kaggle/working/modelisup.h5'\nmodel.save(model_path)","metadata":{"execution":{"iopub.status.busy":"2023-02-22T01:27:10.715669Z","iopub.execute_input":"2023-02-22T01:27:10.716389Z","iopub.status.idle":"2023-02-22T01:28:35.910292Z","shell.execute_reply.started":"2023-02-22T01:27:10.716351Z","shell.execute_reply":"2023-02-22T01:28:35.908597Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from PIL import Image\n\npath = '/kaggle/input/panda-resized-tt/test_images/test_images/0005f7aaab2800f6170c399693a96917.png'\n\nimport matplotlib.pyplot as plt\n\nfig = plt.figure(figsize=(15, 7)) \nrows = 1\ncols = 3\nimg1 = Image.open(path)\nimg2 = img1.convert(\"L\")\n\nfig.add_subplot(rows, cols, 1)\nplt.imshow(img1)\nplt.axis('off')\n\nfig.add_subplot(rows, cols, 2)\nplt.imshow(img2)\nplt.axis('off')\n\nimg2.thumbnail((64,64), Image.ANTIALIAS)\n\nfig.add_subplot(rows, cols, 3)\nplt.imshow(img2) \nplt.axis('off')\n","metadata":{"execution":{"iopub.status.busy":"2023-02-21T19:08:29.638141Z","iopub.execute_input":"2023-02-21T19:08:29.638614Z","iopub.status.idle":"2023-02-21T19:08:29.991673Z","shell.execute_reply.started":"2023-02-21T19:08:29.638578Z","shell.execute_reply":"2023-02-21T19:08:29.990316Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#biopsy severity classification   0 : isup_grade<=1 \n#                                 1 : isup_grade>1\n\nimport numpy as np\nimport os\nfrom PIL import Image\nfrom matplotlib import pyplot as plt\nimport pandas as pd\n\ndef get_data(data_dir):\n    data = [] \n    path = os.path.join(data_dir)\n    for img in os.listdir(path):\n        if img.endswith('.png'):\n            img = Image.open(os.path.join(path, img)).convert(\"L\")\n            img.load()\n            img.thumbnail((64,64), Image.ANTIALIAS)\n            array = np.asarray(img, dtype=\"int32\")\n            data.append(array)                       \n    return np.array(list(data))\n\nX = get_data(\"/kaggle/input/panda-resized-tt/train_images/train_images/train_images\")\nX_test = get_data('/kaggle/input/panda-resized-tt/test_images/test_images')\n\nprint(\"gata x\")\ndf = pd.read_csv(\"/kaggle/input/panda-resized-tt/train1.csv\")\ndf['result'] = np.where(df['isup_grade'].astype(int)<=1, 0, 1)\nY = df['result'].astype(str).to_numpy()\n\ndf1 = pd.read_csv(\"/kaggle/input/panda-resized-tt/test.csv\")\ndf1['result'] = np.where(df1['isup_grade'].astype(int)<=1, 0, 1)\nY_test = df1['result'].astype(str).to_numpy()\n\nfrom tensorflow.keras.utils import to_categorical\n\nX = np.expand_dims(X, axis=3)\nY = to_categorical(Y)\nX_test = np.expand_dims(X_test, axis=3)\nY_test = to_categorical(Y_test)\n\n\nfrom tensorflow import keras\nfrom tensorflow.keras.models import Sequential\nfrom tensorflow.keras.layers import Dense, Dropout, Flatten\nfrom tensorflow.keras.layers import Conv2D, MaxPooling2D\n\nnum_output_classes = 2\ninput_img_size = (64, 64, 1)  # 64x64 image with 1 color channel\n\nmodel = Sequential()\nmodel.add(Conv2D(32, kernel_size=(3, 3), activation=\"relu\", input_shape=input_img_size))\nmodel.add(Conv2D(64, (3, 3), activation=\"relu\"))\nmodel.add(MaxPooling2D(pool_size=(2, 2)))\nmodel.add(Dropout(0.25))\nmodel.add(Flatten())\nmodel.add(Dense(64, activation=\"relu\"))\nmodel.add(Dropout(0.5))\nmodel.add(Dense(num_output_classes, activation=\"softmax\"))\n\nmodel.compile(\n    loss=keras.losses.categorical_crossentropy,\n    optimizer=keras.optimizers.Adadelta(),\n    metrics=[\"accuracy\"],\n)\n\nbatch_size = 128\nepochs = 15\n\nhistory  = model.fit(X, Y,\n          batch_size=batch_size,\n          epochs=epochs,\n          verbose=1,\n          validation_split=0.15)\n\nscore = model.evaluate(X_test, Y_test, verbose=0)\nprint('Test loss:', score[0])\nprint('Test accuracy:', score[1])\nmodel_path ='/kaggle/working/modelB2.h5'\nmodel.save(model_path)","metadata":{"execution":{"iopub.status.busy":"2023-02-21T20:29:38.878506Z","iopub.execute_input":"2023-02-21T20:29:38.878925Z","iopub.status.idle":"2023-02-21T20:54:33.636954Z","shell.execute_reply.started":"2023-02-21T20:29:38.878895Z","shell.execute_reply":"2023-02-21T20:54:33.635469Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#segmentation try\n\nimport tensorflow as tf\nfrom tensorflow import keras\nimport numpy as np\nimport matplotlib.pyplot as plt\nimport openslide\nimport os\nimport pandas as pd\nfrom keras.models import Model\nfrom tensorflow.keras.optimizers import Adam\nimgFolder = '/kaggle/input/panda-resized-train-data-512x512/train_images/train_images'\nmaskFolder = '/kaggle/input/panda-resized-train-data-512x512/train_label_masks/train_label_masks'\nfrom keras.layers import Conv2DTranspose,Conv2D,Dropout,MaxPooling2D, concatenate,UpSampling2D, BatchNormalization\nfrom tensorflow.keras.layers import Activation\nfrom tensorflow.keras.preprocessing.image import load_img\nfrom tensorflow.python.keras.models import Input\nfrom PIL import ImageOps,Image\nimport types\nimport keras_segmentation\n\n# path1='/kaggle/input/panda-resized-train-data-512x512/train_images/train_images/'\n# id1='00a97615a51ba4c475bdec8505623bf9.png'\n# display(Image(filename=path1+id1))\n\n# path2='/kaggle/input/panda-resized-train-data-512x512/train_label_masks/train_label_masks/'\n# id2='00a97615a51ba4c475bdec8505623bf9'\n# id2='00a76bfbec239fd9f465d6581806ff42'\n# end='_mask.png'\n# img = ImageOps.autocontrast(load_img(path2+id2+end))\n# display(img)\nIMG_SHAPE= 128\n\noriginal=[]\nmask=[]\nfor img in os.listdir(maskFolder):\n    mask.append(os.path.join(maskFolder, img))\n    imgId = img[:-9]\n    imgId = imgId+\".png\"\n    original.append(os.path.join(imgFolder, imgId))\ndef scale_down(image, mask):\n  # apply scaling to image and mask\n    image = tf.cast(image, tf.float32) / 255.0\n    mask -= 1\n    return image, mask\n\ndef load_and_preprocess(img_filepath, mask_filepath):\n   # load the image and resize it\n    img = tf.io.read_file(img_filepath)\n    img = tf.io.decode_jpeg(img, channels=3)\n    img = tf.image.resize(img, [IMG_SHAPE, IMG_SHAPE])\n\n    mask = tf.io.read_file(mask_filepath)\n    mask = tf.io.decode_png(mask, channels=1)\n    mask = tf.image.resize(mask, [IMG_SHAPE, IMG_SHAPE])\n\n    img, mask = scale_down(img, mask)\n    return img, mask\n\n\ninput_img_paths_train, mask_img_paths_train = original[: -1000], mask[: -1000]\ninput_img_paths_test, mask_img_paths_test = original[-1000:], mask[-1000:]\n\n\ntrainloader = tf.data.Dataset.from_tensor_slices((input_img_paths_train, mask_img_paths_train))\ntestloader = tf.data.Dataset.from_tensor_slices((input_img_paths_test, mask_img_paths_test))\n\nAUTO = tf.data.experimental.AUTOTUNE\nBATCH_SIZE = 32\n\ntrainloader = (\n    trainloader\n    .shuffle(1024)\n    .map(load_and_preprocess, num_parallel_calls=AUTO)\n    .batch(BATCH_SIZE)\n    .prefetch(AUTO)\n)\n\n\ntestloader = (\n    testloader\n    .map(load_and_preprocess, num_parallel_calls=AUTO)\n    .batch(BATCH_SIZE)\n    .prefetch(AUTO)\n)\n\n\nclass SegmentationModel:\n    def prepare_model(self, OUTPUT_CHANNEL, input_size=(IMG_SHAPE,IMG_SHAPE,3)):\n        inputs = Input(input_size)\n\n\n        # Encoder \n        conv1, pool1 = self.__ConvBlock(32, (3,3), (2,2), 'relu', 'same', inputs) \n        conv2, pool2 = self.__ConvBlock(64, (3,3), (2,2), 'relu', 'same', pool1)\n        conv3, pool3 = self.__ConvBlock(128, (3,3), (2,2), 'relu', 'same', pool2) \n        conv4, pool4 = self.__ConvBlock(256, (3,3), (2,2), 'relu', 'same', pool3) \n\n        # Decoder\n        conv5, up6 = self.__UpConvBlock(512, 256, (3,3), (2,2), (2,2), 'relu', 'same', pool4, conv4)\n        conv6, up7 = self.__UpConvBlock(256, 128, (3,3), (2,2), (2,2), 'relu', 'same', up6, conv3)\n        conv7, up8 = self.__UpConvBlock(128, 64, (3,3), (2,2), (2,2), 'relu', 'same', up7, conv2)\n        conv8, up9 = self.__UpConvBlock(64, 32, (3,3), (2,2), (2,2), 'relu', 'same', up8, conv1)\n\n        conv9 = self.__ConvBlock(32, (3,3), (2,2), 'relu', 'same', up9, False)\n\n        # Notice OUTPUT_CHANNEL and activation\n        outputs = Conv2D(OUTPUT_CHANNEL, (3, 3), activation='softmax', padding='same')(conv9)\n\n\n        return Model(inputs=[inputs], outputs=[outputs])  \n\n\n    def __ConvBlock(self, filters, kernel_size, pool_size, activation, padding, connecting_layer, pool_layer=True):\n        conv = Conv2D(filters=filters, kernel_size=kernel_size, activation=activation, padding=padding)(connecting_layer)\n        conv = Conv2D(filters=filters, kernel_size=kernel_size, activation=activation, padding=padding)(conv)\n        if pool_layer:\n            pool = MaxPooling2D(pool_size)(conv)\n            return conv, pool\n        else:\n            return conv\n\n\n    def __UpConvBlock(self, filters, up_filters, kernel_size, up_kernel, up_stride, activation, padding, connecting_layer, shared_layer):\n        conv = Conv2D(filters=filters, kernel_size=kernel_size, activation=activation, padding=padding)(connecting_layer)\n        conv = Conv2D(filters=filters, kernel_size=kernel_size, activation=activation, padding=padding)(conv)\n        up = Conv2DTranspose(filters=up_filters, kernel_size=up_kernel, strides=up_stride, padding=padding)(conv)\n        up = concatenate([up, shared_layer], axis=3)\n        return conv, up\n\n\nOUTPUT_CHANNEL = 3\n\n\nmodel = SegmentationModel().prepare_model(OUTPUT_CHANNEL)\nmodel.compile(optimizer=\"adam\", loss=\"sparse_categorical_crossentropy\")\n\nsegmentation_classes = ['background', 'benign', 'malign']\n\n\n# returns a dictionary of labels\ndef labels():\n    l = {}\n    for i, label in enumerate(segmentation_classes):\n        l[i] = label\n    return l\n\n\n# util function for generating interactive image mask from components\ndef wandb_mask(bg_img, pred_mask, true_mask):\n      return wandb.Image(bg_img, masks={\n      \"prediction\" : {\n          \"mask_data\" : pred_mask, \n          \"class_labels\" : labels()\n      },\n      \"ground truth\" : {\n          \"mask_data\" : true_mask, \n          \"class_labels\" : labels()\n      }\n    }\n  )\n\nclass SemanticLogger(tf.keras.callbacks.Callback):\n    def __init__(self):\n        super(SemanticLogger, self).__init__()\n        self.val_images, self.val_masks = next(iter(testloader))\n\n\n    def on_epoch_end(self, logs, epoch):\n        pred_masks = self.model.predict(self.val_images)\n        pred_masks = np.argmax(pred_masks, axis=-1)\n        # pred_masks = np.expand_dims(pred_masks, axis=-1)\n\n\n        val_images = tf.image.convert_image_dtype(self.val_images, tf.uint8)\n        val_masks = tf.image.convert_image_dtype(self.val_masks, tf.uint8)\n        val_masks = tf.squeeze(val_masks, axis=-1)\n        \n        pred_masks = tf.image.convert_image_dtype(pred_masks, tf.uint8)\n\n\n        mask_list = []\n        for i in range(len(self.val_images)):\n            mask_list.append(wandb_mask(val_images[i].numpy(), \n                                      pred_masks[i].numpy(), \n                                      val_masks[i].numpy()))\n\n\n        wandb.log({\"predictions\" : mask_list})\n\ncallbacks = [\n   SemanticLoger()\n]\nepochs = 15\nmodel.fit(trainloader, epochs=epochs, validation_data=testloader, callbacks=callbacks)\n\n\n\n\n# import matplotlib.image\n# matplotlib.image.imsave('/kaggle/working/im.png', np.array([pred])*255.0)\n\n# display(ImageOps.autocontrast(load_img(\"/kaggle/working/im.png\")))","metadata":{"execution":{"iopub.status.busy":"2023-02-21T09:08:18.779628Z","iopub.execute_input":"2023-02-21T09:08:18.780762Z","iopub.status.idle":"2023-02-21T09:08:19.719927Z","shell.execute_reply.started":"2023-02-21T09:08:18.780687Z","shell.execute_reply":"2023-02-21T09:08:19.718482Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import cv2\nimport os\nimport pandas as pd\nimport numpy as np\nfrom matplotlib import pyplot as plt\nfrom keras.models import Sequential\nfrom keras.layers import Flatten, Conv2D, MaxPool2D, Activation, Dense, Dropout\nfrom keras.optimizers import Adam\nfrom keras.preprocessing.image import ImageDataGenerator\nfrom keras.applications.vgg16 import preprocess_input\nfrom tensorflow.keras.utils import load_img,  img_to_array\nimport tensorflow as tf\nfrom keras.optimizers import SGD\n\ndf = pd.read_csv(\"/kaggle/input/prostate-cancer-grade-assessment/train.csv\")\ndata_dir = '/kaggle/input/panda-resized-train-data-512x512/train_images/'\npath ='/kaggle/input/panda-resized-train-data-512x512/train_images/train_images'\nmodel_path ='/kaggle/working/modelBinar7.h5'\n\ndf['result'] = np.where(df['isup_grade'].astype(int)<=1, 0, 1)\ndf['image_id'] = df['image_id'].astype(str) + '.png' \n\n# train_datagen = ImageDataGenerator(\n#     rescale=1./255,\n#     validation_split=0.05,\n# #     shear_range=0.2,\n# #     zoom_range=0.2,\n# #     horizontal_flip=True\n# )\n# df['result'] = df['result'].astype(str)\n\n# train_generator = train_datagen.flow_from_dataframe(\n#     df,\n#     directory = path,\n#     subset = 'training',\n#     x_col = 'image_id',\n#     y_col = 'result',\n#     class_mode = 'binary'\n# )\n# val_generator = train_datagen.flow_from_dataframe(\n#     df,\n#     directory = path,\n#     subset = 'validation',\n#     x_col = 'image_id',\n#     y_col = 'result',\n#     class_mode = 'binary'\n# )\n  \nmodel = Sequential()\nmodel.add(Conv2D(32, (3,3) ,activation = 'relu', input_shape = (256,256,3)))\nmodel.add(Conv2D(32, (3,3), activation = 'relu'))\nmodel.add(MaxPool2D(2,2))\nmodel.add(Conv2D(64, (3,3), activation='relu'))\nmodel.add(Conv2D(64, (3,3), activation='relu'))\nmodel.add(MaxPool2D(2,2))\n# model.add(Conv2D(128, (3,3), activation='relu'))\n# model.add(MaxPool2D(2,2))\nmodel.add(Flatten())\nmodel.add(Dense(512, activation = 'relu'))\nmodel.add(Dropout(0.2))\nmodel.add(Dense(1, activation = 'sigmoid'))    \n    \nmodel.compile(\n    loss = 'binary_crossentropy',\n    optimizer = SGD(lr=0.01), \n    metrics = ['accuracy']\n)\n\nmodel.summary()\n\nhistory = model.fit(\n    train_generator,\n    steps_per_epoch = 20,\n    epochs = 1,\n    validation_data = val_generator,\n    validation_steps = 5\n)\n\nmy_image = load_img('/kaggle/input/panda-resized-train-data-512x512/train_images/train_images/007433133235efc27a39f11df6940829.png', target_size=(256, 256))\nmy_image = img_to_array(my_image)\nmy_image = my_image.reshape((1, my_image.shape[0], my_image.shape[1], my_image.shape[2]))\nprediction = model.predict(my_image)\nprint([np.round(x) for x in prediction])\n\nmi = load_img('/kaggle/input/panda-resized-train-data-512x512/train_images/train_images/0005f7aaab2800f6170c399693a96917.png', target_size=(256, 256))\nmi = img_to_array(mi)\nmi = mi.reshape((1, mi.shape[0], mi.shape[1], mi.shape[2]))\nprediction = model.predict(mi)\nprint([np.round(x) for x in prediction])\n\nmi = load_img('/kaggle/input/panda-resized-train-data-512x512/train_images/train_images/0018ae58b01bdadc8e347995b69f99aa.png', target_size=(256, 256))\nmi = img_to_array(mi)\nmi = mi.reshape((1, mi.shape[0], mi.shape[1], mi.shape[2]))\nprediction = model.predict(mi)\nprint([np.round(x) for x in prediction])\n# model.save(model_path)","metadata":{"execution":{"iopub.status.busy":"2023-02-18T23:35:57.357957Z","iopub.execute_input":"2023-02-18T23:35:57.358443Z","iopub.status.idle":"2023-02-18T23:38:22.213513Z","shell.execute_reply.started":"2023-02-18T23:35:57.358408Z","shell.execute_reply":"2023-02-18T23:38:22.212075Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from tqdm import tqdm\nimport tensorflow as tf\nimport numpy as np\nimport os\nfrom PIL import Image\nfrom matplotlib import pyplot as plt\nimport pandas as pd\n\ndef get_data(data_dir):\n    data = [] \n    path = os.path.join(data_dir)\n    for img in os.listdir(path):\n        img = Image.open(os.path.join(path, img)).convert(\"L\")\n        img.load()\n        img.thumbnail((64,64), Image.ANTIALIAS)\n        array = np.asarray(img, dtype=\"int32\")\n        data.append(array)                       \n    return np.array(list(data))\n\nX = get_data(\"/kaggle/input/panda-resized-train-data-512x512/train_images/train_images\")\n\nprint(\"gata x\")\n\ndf = pd.read_csv(\"/kaggle/input/prostate-cancer-grade-assessment/train.csv\")\ndf['result'] = np.where(df['isup_grade'].astype(int)<=1, 0, 1)\n\nY = df['result'].astype(str).to_numpy()\n\nfrom tensorflow.keras.utils import to_categorical\nX = np.expand_dims(X, axis=3)\nY = to_categorical(Y)\n\nfrom tensorflow import keras\n\nfrom tensorflow.keras.models import Sequential\nfrom tensorflow.keras.layers import Dense, Dropout, Flatten\nfrom tensorflow.keras.layers import Conv2D, MaxPooling2D\n\nnum_output_classes = 2\ninput_img_size = (64, 64, 1)  # 64x64 image with 1 color channel\n\nmodel = Sequential()\nmodel.add(Conv2D(32, kernel_size=(3, 3), activation=\"relu\", input_shape=input_img_size))\nmodel.add(Conv2D(64, (3, 3), activation=\"relu\"))\nmodel.add(MaxPooling2D(pool_size=(2, 2)))\nmodel.add(Dropout(0.25))\nmodel.add(Flatten())\nmodel.add(Dense(64, activation=\"relu\"))\nmodel.add(Dropout(0.5))\nmodel.add(Dense(num_output_classes, activation=\"softmax\"))\nmodel.compile(\n    loss=keras.losses.categorical_crossentropy,\n    optimizer=keras.optimizers.Adadelta(),\n    metrics=[\"accuracy\"],\n)\n\nbatch_size = 128\nepochs = 2\n\nhistory  = model.fit(X, Y,\n          batch_size=batch_size,\n          epochs=epochs,\n          verbose=1,\n          validation_split=0.2)\n\n# img = Image.open('/kaggle/input/panda-resized-train-data-512x512/train_images/train_images/007433133235efc27a39f11df6940829.png').convert(\"L\")\n# img.load()\n# img.thumbnail((64,64), Image.ANTIALIAS)\n# mi = np.asarray(img, dtype=\"int32\")\n# mi = np.expand_dims(mi, axis=0)\n# mi = np.expand_dims(mi, axis=3)\n# prediction = model.predict(mi)\n# print([x for x in prediction])\n\n# img = Image.open('/kaggle/input/panda-resized-train-data-512x512/train_images/train_images/0005f7aaab2800f6170c399693a96917.png').convert(\"L\")\n# img.load()\n# img.thumbnail((64,64), Image.ANTIALIAS)\n# mi = np.asarray(img, dtype=\"int32\")\n# mi = np.expand_dims(mi, axis=0)\n# mi = np.expand_dims(mi, axis=3)\n# prediction = model.predict(mi)\n# print([x for x in prediction])\n\n# img = Image.open('/kaggle/input/panda-resized-train-data-512x512/train_images/train_images/0018ae58b01bdadc8e347995b69f99aa.png').convert(\"L\")\n# img.load()\n# img.thumbnail((64,64), Image.ANTIALIAS)\n# mi = np.asarray(img, dtype=\"int32\")\n# mi = np.expand_dims(mi, axis=0)\n# mi = np.expand_dims(mi, axis=3)\n# prediction = model.predict(mi)\n# print([x for x in prediction])\n\n# model_path ='/kaggle/working/modelB.h5'\n# model.save(model_path)","metadata":{"execution":{"iopub.status.busy":"2023-02-20T22:09:50.379719Z","iopub.execute_input":"2023-02-20T22:09:50.38006Z","iopub.status.idle":"2023-02-20T22:09:50.513976Z","shell.execute_reply.started":"2023-02-20T22:09:50.380034Z","shell.execute_reply":"2023-02-20T22:09:50.512946Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import visualkeras\nvisualkeras.layered_view(model)","metadata":{"execution":{"iopub.status.busy":"2023-02-20T22:09:23.551345Z","iopub.execute_input":"2023-02-20T22:09:23.551689Z","iopub.status.idle":"2023-02-20T22:09:34.452492Z","shell.execute_reply.started":"2023-02-20T22:09:23.551663Z","shell.execute_reply":"2023-02-20T22:09:34.451444Z"},"trusted":true},"execution_count":null,"outputs":[]}],"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}}