{"metadata":{"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"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"!pip install efficientnet -q ","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import os \n\nimport efficientnet.tfkeras as efn \nimport numpy as np\nimport pandas as pd\n\n\nfrom kaggle_datasets import KaggleDatasets\nfrom sklearn.model_selection import train_test_split\nimport tensorflow as tf \nfrom sklearn.model_selection import GroupKFold ","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def auto_select_accelerator():\n    try:\n        tpu = tf.distribute.cluster_resolver.TPUClusterResolver()\n        tf.config.experimental_connect_to_cluster(tpu)\n        tf.tpu.experimental.initialize_tpu_system(tpu)\n        strategy = tf.distribute.experimental.TPUStrategy(tpu)\n        print(\"Running on TPU:\", tpu.master())\n    except ValueError:\n        strategy = tf.distribute.get_strategy()\n    print(f\"Running on {strategy.num_replicas_in_sync} replicas\")\n    \n    return strategy\n\n\ndef build_decoder(with_labels=True, target_size=(256, 256), ext='jpg'):\n    def decode(path):\n        file_bytes = tf.io.read_file(path)\n\n        if ext == 'png':\n            img = tf.image.decode_png(file_bytes, channels=3)\n        elif ext in ['jpg', 'jpeg']:\n            img = tf.image.decode_jpeg(file_bytes, channels=3)\n        else:\n            raise ValueError(\"Image extension not supported\")\n        img = tf.cast(img, tf.float32) / 255.0\n        img = tf.image.resize(img, target_size)\n\n        return img\n    \n    def decode_with_labels(path, label):\n        return decode(path), label\n    \n    return decode_with_labels if with_labels else decode\n\n\ndef build_augmenter(with_labels=True):\n    def augment(img):\n        img = tf.image.random_flip_left_right(img)\n        img = tf.image.random_flip_up_down(img)\n        return img\n    \n    def augment_with_labels(img, label):\n        return augment(img), label\n    \n    return augment_with_labels if with_labels else augment\n\n\ndef build_dataset(paths, labels=None, bsize=128, cache=True,\n                  decode_fn=None, augment_fn=None,\n                  augment=True, repeat=True, shuffle=1024, \n                  cache_dir=\"\"):\n    if cache_dir != \"\" and cache is True:\n        os.makedirs(cache_dir, exist_ok=True)\n    \n    if decode_fn is None:\n        decode_fn = build_decoder(labels is not None)\n    \n    if augment_fn is None:\n        augment_fn = build_augmenter(labels is not None)\n    \n    AUTO = tf.data.experimental.AUTOTUNE\n    slices = paths if labels is None else (paths, labels)\n    \n    dset = tf.data.Dataset.from_tensor_slices(slices)\n    dset = dset.map(decode_fn, num_parallel_calls=AUTO)\n    dset = dset.cache(cache_dir) if cache else dset\n    dset = dset.map(augment_fn, num_parallel_calls=AUTO) if augment else dset\n    dset = dset.repeat() if repeat else dset\n    dset = dset.shuffle(shuffle) if shuffle else dset\n    dset = dset.batch(bsize).prefetch(AUTO)\n    \n    return dset","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"COMPETITION_NAME = 'hpa-768768'\nstrategy = auto_select_accelerator()\nBATCH_SIZE = strategy.num_replicas_in_sync * 16\nGCS_DS_PATH = KaggleDatasets().get_gcs_path(COMPETITION_NAME)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"print(GCS_DS_PATH)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"load_dir = f\"/kaggle/input/{COMPETITION_NAME}/\"\ndf = pd.read_csv('../input/classification-label-csv-green/df_green.csv')\nlabel_cols = df.columns[2:21]\npaths = GCS_DS_PATH + '/' + df['ID'] + '.png'\nlabels = df[label_cols].values","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_paths, valid_paths, train_labels, valid_labels = train_test_split(paths, labels, test_size = 0.2, random_state = 21)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"IMSIZE = (224, 240, 260, 300, 380, 456, 528, 600)\nIMS = 7 \n\ndecoder = build_decoder(with_labels=True, target_size=(IMSIZE[IMS], IMSIZE[IMS]))\ntest_decoder = build_decoder(with_labels=False, target_size=(IMSIZE[IMS], IMSIZE[IMS]))\n\ntrain_dataset = build_dataset(\n    train_paths, train_labels, bsize=BATCH_SIZE, decode_fn=decoder\n)\n\nvalid_dataset = build_dataset(\n    valid_paths, valid_labels, bsize=BATCH_SIZE, decode_fn=decoder,\n    repeat=False, shuffle=False, augment=False\n)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"try:\n    n_labels = train_labels.shape[1]\nexcept:\n    n_labels = 1\n    \nwith strategy.scope():\n    model = tf.keras.Sequential([\n        efn.EfficientNetB7(\n        input_shape = (IMSIZE[IMS], IMSIZE[IMS], 3),\n            weights = 'imagenet',\n            include_top = False\n        ),\n        tf.keras.layers.GlobalAveragePooling2D(),\n        tf.keras.layers.Dense(n_labels, activation = 'sigmoid')\n    ])\n    \n    model.compile(\n    optimizer = tf.keras.optimizers.Adam(),\n        loss = 'binary_crossentropy',\n        metrics = [tf.keras.metrics.AUC(multi_label = True)]\n    )\n    model.summary()","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"color = '_green'\n\n\nsteps_per_epoch = train_paths.shape[0] // BATCH_SIZE\n\ncheckpoint = tf.keras.callbacks.ModelCheckpoint( \n    f'model{color}.h5', save_best_only = True, monitor = 'val_loss', mode = 'min'\n)\nlr_reducer = tf.keras.callbacks.ReduceLROnPlateau( \n    monitor = 'val_loss', patience = 3, min_lr = 1e-6, mode = 'min'\n)\n\n","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"history = model.fit(train_dataset, epochs = 20, verbose = 1, callbacks = [checkpoint, lr_reducer], steps_per_epoch = steps_per_epoch, validation_data = valid_dataset)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"hist_df = pd.DataFrame(history.history)\nhist_df.to_csv(f'history{color}_effb7.csv')","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"try:\n    n_labels = train_labels.shape[1]\nexcept:\n    n_labels = 1\n    \nwith strategy.scope():\n    model = tf.keras.Sequential([\n        efn.EfficientNetB4(\n        input_shape = (IMSIZE[IMS], IMSIZE[IMS], 3),\n            weights = 'imagenet',\n            include_top = False\n        ),\n        tf.keras.layers.GlobalAveragePooling2D(),\n        tf.keras.layers.Dense(n_labels, activation = 'sigmoid')\n    ])\n    \n    model.compile(\n    optimizer = tf.keras.optimizers.Adam(),\n        loss = 'binary_crossentropy',\n        metrics = [tf.keras.metrics.AUC(multi_label = True)]\n    )\n    model.summary()\n    \ncolor = '_green'\n\n\nsteps_per_epoch = train_paths.shape[0] // BATCH_SIZE\n\ncheckpoint = tf.keras.callbacks.ModelCheckpoint( \n    f'model{color}_b4.h5', save_best_only = True, monitor = 'val_loss', mode = 'min'\n)\nlr_reducer = tf.keras.callbacks.ReduceLROnPlateau( \n    monitor = 'val_loss', patience = 3, min_lr = 1e-6, mode = 'min'\n)\nhistory = model.fit(train_dataset, epochs = 20, verbose = 1, callbacks = [checkpoint, lr_reducer], steps_per_epoch = steps_per_epoch, validation_data = valid_dataset)\n\n\nhist_df = pd.DataFrame(history.history)\nhist_df.to_csv(f'history{color}_effb4.csv')","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"try:\n    n_labels = train_labels.shape[1]\nexcept:\n    n_labels = 1\n    \nwith strategy.scope():\n    model = tf.keras.Sequential([\n        efn.EfficientNetB1(\n        input_shape = (IMSIZE[IMS], IMSIZE[IMS], 3),\n            weights = 'imagenet',\n            include_top = False\n        ),\n        tf.keras.layers.GlobalAveragePooling2D(),\n        tf.keras.layers.Dense(n_labels, activation = 'sigmoid')\n    ])\n    \n    model.compile(\n    optimizer = tf.keras.optimizers.Adam(),\n        loss = 'binary_crossentropy',\n        metrics = [tf.keras.metrics.AUC(multi_label = True)]\n    )\n    model.summary()\n    \ncolor = '_green'\n\n\nsteps_per_epoch = train_paths.shape[0] // BATCH_SIZE\n\ncheckpoint = tf.keras.callbacks.ModelCheckpoint( \n    f'model{color}_b1.h5', save_best_only = True, monitor = 'val_loss', mode = 'min'\n)\nlr_reducer = tf.keras.callbacks.ReduceLROnPlateau( \n    monitor = 'val_loss', patience = 3, min_lr = 1e-6, mode = 'min'\n)\n\nhistory = model.fit(train_dataset, epochs = 20, verbose = 1, callbacks = [checkpoint, lr_reducer], steps_per_epoch = steps_per_epoch, validation_data = valid_dataset)\n\nhist_df = pd.DataFrame(history.history)\nhist_df.to_csv(f'history{color}_effb1.csv')","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Modified Training","metadata":{}},{"cell_type":"code","source":"!pip install -q efficientnet \n\n!pip install focal-loss ","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import os \n\nimport efficientnet.tfkeras as efn \nimport numpy as np \nimport pandas as pd \nimport tensorflow as tf \n\nfrom focal_loss import BinaryFocalLoss \nfrom kaggle_datasets import KaggleDatasets \nfrom sklearn.model_selection import train_test_split \nfrom sklearn.model_selection import GroupKFold \n\n","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def auto_select_accelerator():\n    try:\n        tpu = tf.distribute.cluster_resolver.TPUClusterResolver()\n        tf.config.experimental_connect_to_cluster(tpu)\n        tf.tpu.experimental.initialize_tpu_system(tpu)\n        strategy = tf.distribute.experimental.TPUStrategy(tpu)\n        print(\"Running on TPU:\", tpu.master())\n    except ValueError:\n        strategy = tf.distribute.get_strategy()\n    print(f\"Running on {strategy.num_replicas_in_sync} replicas\")\n    \n    return strategy\n\n\ndef build_decoder(with_labels=True, target_size=(256, 256), ext='jpg'):\n    def decode(path):\n        file_bytes = tf.io.read_file(path)\n\n        if ext == 'png':\n            img = tf.image.decode_png(file_bytes, channels=3)\n        elif ext in ['jpg', 'jpeg']:\n            img = tf.image.decode_jpeg(file_bytes, channels=3)\n        else:\n            raise ValueError(\"Image extension not supported\")\n        img = tf.cast(img, tf.float32) / 255.0\n        img = tf.image.resize(img, target_size)\n\n        return img\n    \n    def decode_with_labels(path, label):\n        return decode(path), label\n    \n    return decode_with_labels if with_labels else decode\n\n\ndef build_augmenter(with_labels=True):\n    def augment(img):\n        img = tf.image.random_flip_left_right(img)\n        img = tf.image.random_flip_up_down(img)\n        return img\n    \n    def augment_with_labels(img, label):\n        return augment(img), label\n    \n    return augment_with_labels if with_labels else augment\n\n\ndef build_dataset(paths, labels=None, bsize=128, cache=True,\n                  decode_fn=None, augment_fn=None,\n                  augment=True, repeat=True, shuffle=1024, \n                  cache_dir=\"\"):\n    if cache_dir != \"\" and cache is True:\n        os.makedirs(cache_dir, exist_ok=True)\n    \n    if decode_fn is None:\n        decode_fn = build_decoder(labels is not None)\n    \n    if augment_fn is None:\n        augment_fn = build_augmenter(labels is not None)\n    \n    AUTO = tf.data.experimental.AUTOTUNE\n    slices = paths if labels is None else (paths, labels)\n    \n    dset = tf.data.Dataset.from_tensor_slices(slices)\n    dset = dset.map(decode_fn, num_parallel_calls=AUTO)\n    dset = dset.cache(cache_dir) if cache else dset\n    dset = dset.map(augment_fn, num_parallel_calls=AUTO) if augment else dset\n    dset = dset.repeat() if repeat else dset\n    dset = dset.shuffle(shuffle) if shuffle else dset\n    dset = dset.batch(bsize).prefetch(AUTO)\n    \n    return dset","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"COMPETITION_NAME = \"hpa-768768\"\nEPOCHS = 20\n\nstrategy = auto_select_accelerator()\nBATCH_SIZE = strategy.num_replicas_in_sync * 20\n\nGCS_DS_PATH = KaggleDatasets().get_gcs_path(COMPETITION_NAME); print(GCS_DS_PATH)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# green\n\nload_dir = f\"/kaggle/input/{COMPETITION_NAME}/\"\n\ndf = pd.read_csv(\"../input/classification-label-csv-green/df_green.csv\")\n\nlabel_cols = df.columns[2: 21]\n\npaths = GCS_DS_PATH + '/' + df['ID'] + '.png'\n\nlabels = df[label_cols].values\n\n\n","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"(\ntrain_paths, valid_paths, \n    train_labels, valid_labels\n) = train_test_split(paths, labels, test_size = 0.2, random_state = 42)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"IMSIZE = (224, 240, 260, 300, 380, 456, 512, 528, 600, 720)\nIMS = -1\n\ndecoder = build_decoder(with_labels = True, target_size = (IMSIZE[IMS], IMSIZE[IMS]))\ntest_decoder = build_decoder(with_labels = False, target_size = (IMSIZE[IMS], IMSIZE[IMS]))\n\ntrain_dataset = build_dataset(train_paths, train_labels, bsize = BATCH_SIZE, decode_fn = decoder)\n\nvalid_dataset = build_dataset(valid_paths, valid_labels, bsize = BATCH_SIZE, decode_fn = decoder, repeat = False, shuffle = False, augment = False)\n\n","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"try:\n    \n    n_labels = train_labels.shape[1]\n    \nexcept:\n    n_labels = 1\n    \n    \nbinary_focal_loss = BinaryFocalLoss(gamma = 2)\n\n\nwith strategy.scope():\n    \n    model = tf.keras.Sequential([\n        efn.EfficientNetB6( \n        input_shape = (IMSIZE[IMS], IMSIZE[IMS], 3),\n            weights = \"imagenet\",\n            include_top = False\n        ),\n        tf.keras.layers.GlobalAveragePooling2D(),\n        tf.keras.layers.Dense(n_labels, activation = 'sigmoid')\n    ])\n    \n    \n    model.compile(\n    optimizer = tf.keras.optimizers.Adam(),\n    loss = 'binary_crossentropy',\n    metrics = [tf.keras.metrics.AUC(multi_label = True)])\n    \n    model.summary()","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"colour = '_green'","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"steps_per_epoch = train_paths.shape[0]//BATCH_SIZE\n\ncheckpoint = tf.keras.callbacks.ModelCheckpoint(\nf'model{colour}_bce.h5', save_best_only = True, monitor = 'val_loss', mode = 'min'\n)\n\nlr_reducer = tf.keras.callbacks.ReduceLROnPlateau(\nmonitor = 'val_loss', patience = 3, min_lr = 1e-6, mode = 'min'\n)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"history = model.fit(\n\ntrain_dataset, \n    epochs = EPOCHS, \n    verbose = 1,\n    callbacks = [checkpoint, lr_reducer],\n    steps_per_epoch = steps_per_epoch,\n    validation_data = valid_dataset\n)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"hist_df = pd.DataFrame(history.history)\n\nhist_df.to_csv(f'history{colour}_bce.csv')","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"try:\n    \n    n_labels = train_labels.shape[1]\n    \nexcept:\n    n_labels = 1\n    \n    \nbinary_focal_loss = BinaryFocalLoss(gamma = 2)\n\n\nwith strategy.scope():\n    \n    model = tf.keras.Sequential([\n        efn.EfficientNetB6( \n        input_shape = (IMSIZE[IMS], IMSIZE[IMS], 3),\n            weights = \"imagenet\",\n            include_top = False\n        ),\n        tf.keras.layers.GlobalAveragePooling2D(),\n        tf.keras.layers.Dense(n_labels, activation = 'sigmoid')\n    ])\n    \n    \n    model.compile(\n    optimizer = tf.keras.optimizers.Adam(),\n    loss = binary_focal_loss,\n    metrics = [tf.keras.metrics.AUC(multi_label = True)])\n    \n    model.summary()\n    \n    \nsteps_per_epoch = train_paths.shape[0]//BATCH_SIZE\n\ncheckpoint = tf.keras.callbacks.ModelCheckpoint(\nf'model{colour}_bfl.h5', save_best_only = True, monitor = 'val_loss', mode = 'min'\n)\n\nlr_reducer = tf.keras.callbacks.ReduceLROnPlateau(\nmonitor = 'val_loss', patience = 3, min_lr = 1e-6, mode = 'min'\n)\n\nhistory = model.fit(\n\ntrain_dataset, \n    epochs = EPOCHS, \n    verbose = 1,\n    callbacks = [checkpoint, lr_reducer],\n    steps_per_epoch = steps_per_epoch,\n    validation_data = valid_dataset\n)\n\nhist_df = pd.DataFrame(history.history)\n\nhist_df.to_csv(f'history{colour}_bfl.csv')","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]}]}