{"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":"markdown","source":"# Code copied with minimal changes from this amazing Notebook:\n\nhttps://www.kaggle.com/jpbremer/backfins-arcface-tpu-effnet\n\n## I try use horizontal flip image and random rotate image (-10, 10 ) degree\n## *** In this notebook, i dont training","metadata":{"execution":{"iopub.status.busy":"2022-03-14T10:40:27.280354Z","iopub.execute_input":"2022-03-14T10:40:27.280562Z","iopub.status.idle":"2022-03-14T10:40:27.293355Z","shell.execute_reply.started":"2022-03-14T10:40:27.280539Z","shell.execute_reply":"2022-03-14T10:40:27.292509Z"}}},{"cell_type":"code","source":"import os","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import tensorflow as tf\ntry:\n    # TPU detection. No parameters necessary if TPU_NAME environment variable is\n    # set: this is always the case on Kaggle.\n    tpu = tf.distribute.cluster_resolver.TPUClusterResolver()\n    print('Running on TPU ', tpu.master())\nexcept ValueError:\n    tpu = None\n\nif tpu:\n    tf.config.experimental_connect_to_cluster(tpu)\n    tf.tpu.experimental.initialize_tpu_system(tpu)\n    strategy = tf.distribute.TPUStrategy(tpu)\nelse:\n    # Default distribution strategy in Tensorflow. Works on CPU and single GPU.\n    strategy = tf.distribute.get_strategy()\n\nAUTO = tf.data.experimental.AUTOTUNE\nprint(\"REPLICAS: \", strategy.num_replicas_in_sync)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from kaggle_datasets import KaggleDatasets","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"!pip install -q efficientnet\n!pip install tensorflow_addons\nimport re\nimport os\nimport numpy as np\nimport pandas as pd\nimport random\nimport math\nimport tensorflow as tf\nimport efficientnet.tfkeras as efn\nfrom sklearn import metrics\nfrom sklearn.model_selection import KFold, train_test_split\nfrom tensorflow.keras import backend as K\nimport tensorflow_addons as tfa\nfrom tqdm.auto import tqdm\nimport matplotlib.pyplot as plt\nimport pickle\nimport json\nimport tensorflow_hub as tfhub\nfrom datetime import datetime","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Config","metadata":{}},{"cell_type":"code","source":"save_dir = '.'\nEXPERIMENT = 0\nrun_ts = datetime.now().strftime('%Y%m%d-%H%M%S')\nprint(run_ts)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"class config:\n    \n    \n    SEED = 1000\n    FOLD_TO_RUN = 0\n    FOLDS = 5\n    DEBUG = False\n    EVALUATE = True\n    RESUME = False\n    RESUME_EPOCH = None\n    \n    \n    ### Dataset\n    BATCH_SIZE = 16 * strategy.num_replicas_in_sync\n    IMAGE_SIZE = 640\n    N_CLASSES = 15587\n    \n    ### Model\n    model_type = 'effnetv1'  \n    EFF_NET = 5\n    EFF_NETV2 = 's-21k-ft1k'\n    FREEZE_BATCH_NORM = False\n    head = 'arcface' \n    EPOCHS = 30\n    LR = 0.001\n    message='baseline'\n    \n    ### Augmentations\n    CUTOUT = False\n    \n    ### Save-Directory\n    save_dir = save_dir\n    \n    ### Inference\n    KNN = 50\n    \ndef count_data_items(filenames):\n    n = [int(re.compile(r\"-([0-9]*)\\.\").search(filename).group(1)) \n         for filename in filenames]\n    return np.sum(n)\n\n# Function to seed everything\ndef seed_everything(seed):\n    random.seed(seed)\n    np.random.seed(seed)\n    os.environ['PYTHONHASHSEED'] = str(seed)\n    tf.random.set_seed(seed)\n    \ndef is_interactive():\n    return 'runtime'    in get_ipython().config.IPKernelApp.connection_file\nIS_INTERACTIVE = is_interactive()\nprint(IS_INTERACTIVE)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"MODEL_NAME = None\nif config.model_type == 'effnetv1':\n    MODEL_NAME = f'effnetv1_b{config.EFF_NET}'\nelif config.model_type == 'effnetv2':\n    MODEL_NAME = f'effnetv2_{config.EFF_NETV2}'\n\nconfig.MODEL_NAME = MODEL_NAME\nprint(MODEL_NAME)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"with open(config.save_dir+'/config.json', 'w') as fp:\n    json.dump({x:dict(config.__dict__)[x] for x in dict(config.__dict__) if not x.startswith('_')}, fp)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"GCS_PATH = KaggleDatasets().get_gcs_path('backfintfrecords')","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_files = np.sort(np.array(tf.io.gfile.glob(GCS_PATH + '/happywhale-2022-train*.tfrec')))\ntest_files = np.sort(np.array(tf.io.gfile.glob(GCS_PATH + '/happywhale-2022-test*.tfrec')))\nprint(GCS_PATH)\nprint(len(train_files),len(test_files),count_data_items(train_files),count_data_items(test_files))","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Data","metadata":{}},{"cell_type":"code","source":"import random","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"random.uniform(-10, 10)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def arcface_format(posting_id, image, label_group, matches):\n    return posting_id, {'inp1': image, 'inp2': label_group}, label_group, matches\n\ndef arcface_inference_format(posting_id, image, label_group, matches):\n    return image,posting_id\n\ndef arcface_eval_format(posting_id, image, label_group, matches):\n    return image,label_group\n\n# Data augmentation function\ndef data_augment(posting_id, image, label_group, matches):\n\n    ### CUTOUT\n    if tf.random.uniform([])>0.5 and config.CUTOUT:\n      N_CUTOUT = 6\n      for cutouts in range(N_CUTOUT):\n        if tf.random.uniform([])>0.5:\n           DIM = config.IMAGE_SIZE\n           CUTOUT_LENGTH = DIM//8\n           x1 = tf.cast( tf.random.uniform([],0,DIM-CUTOUT_LENGTH),tf.int32)\n           x2 = tf.cast( tf.random.uniform([],0,DIM-CUTOUT_LENGTH),tf.int32)\n           filter_ = tf.concat([tf.zeros((x1,CUTOUT_LENGTH)),tf.ones((CUTOUT_LENGTH,CUTOUT_LENGTH)),tf.zeros((DIM-x1-CUTOUT_LENGTH,CUTOUT_LENGTH))],axis=0)\n           filter_ = tf.concat([tf.zeros((DIM,x2)),filter_,tf.zeros((DIM,DIM-x2-CUTOUT_LENGTH))],axis=1)\n           cutout = tf.reshape(1-filter_,(DIM,DIM,1))\n           image = cutout*image\n\n    image = tf.image.random_flip_left_right(image)\n    # image = tf.image.random_flip_up_down(image)\n#     degree = random.uniform(-10, 10)\n#     image = tfa.image.rotate(image, degree * math.pi / 180)\n    \n    image = tf.image.random_hue(image, 0.01)\n    image = tf.image.random_saturation(image, 0.70, 1.30)\n    image = tf.image.random_contrast(image, 0.80, 1.20)\n    image = tf.image.random_brightness(image, 0.10)\n    return posting_id, image, label_group, matches\n\n# Function to decode our images\ndef decode_image(image_data):\n    image = tf.image.decode_jpeg(image_data, channels = 3)\n    image = tf.image.resize(image, [config.IMAGE_SIZE,config.IMAGE_SIZE])\n    image = tf.cast(image, tf.float32) / 255.0\n    return image\n\n# This function parse our images and also get the target variable\ndef read_labeled_tfrecord(example):\n    LABELED_TFREC_FORMAT = {\n        \"image_name\": tf.io.FixedLenFeature([], tf.string),\n        \"image\": tf.io.FixedLenFeature([], tf.string),\n        \"target\": tf.io.FixedLenFeature([], tf.int64),\n#         \"matches\": tf.io.FixedLenFeature([], tf.string)\n    }\n\n    example = tf.io.parse_single_example(example, LABELED_TFREC_FORMAT)\n    posting_id = example['image_name']\n    image = decode_image(example['image'])\n#     label_group = tf.one_hot(tf.cast(example['label_group'], tf.int32), depth = N_CLASSES)\n    label_group = tf.cast(example['target'], tf.int32)\n#     matches = example['matches']\n    matches = 1\n    return posting_id, image, label_group, matches\n\n# This function loads TF Records and parse them into tensors\ndef load_dataset(filenames, ordered = False):\n    \n    ignore_order = tf.data.Options()\n    if not ordered:\n        ignore_order.experimental_deterministic = False \n        \n    dataset = tf.data.TFRecordDataset(filenames, num_parallel_reads = AUTO)\n#     dataset = dataset.cache()\n    dataset = dataset.with_options(ignore_order)\n    dataset = dataset.map(read_labeled_tfrecord, num_parallel_calls = AUTO) \n    return dataset\n\n# This function is to get our training tensors\ndef get_training_dataset(filenames):\n    dataset = load_dataset(filenames, ordered = False)\n    dataset = dataset.map(data_augment, num_parallel_calls = AUTO)\n    dataset = dataset.map(arcface_format, num_parallel_calls = AUTO)\n    dataset = dataset.map(lambda posting_id, image, label_group, matches: (image, label_group))\n    dataset = dataset.repeat()\n    dataset = dataset.shuffle(2048)\n    dataset = dataset.batch(config.BATCH_SIZE)\n    dataset = dataset.prefetch(AUTO)\n    return dataset\n\n# This function is to get our training tensors\ndef get_val_dataset(filenames):\n    dataset = load_dataset(filenames, ordered = True)\n    dataset = dataset.map(data_augment, num_parallel_calls = AUTO)\n    dataset = dataset.map(arcface_format, num_parallel_calls = AUTO)\n    dataset = dataset.map(lambda posting_id, image, label_group, matches: (image, label_group))\n    dataset = dataset.batch(config.BATCH_SIZE)\n    dataset = dataset.prefetch(AUTO)\n    return dataset\n\n# This function is to get our training tensors\ndef get_eval_dataset(filenames, get_targets = True):\n    dataset = load_dataset(filenames, ordered = True)\n    dataset = dataset.map(data_augment, num_parallel_calls = AUTO)\n    dataset = dataset.map(arcface_eval_format, num_parallel_calls = AUTO)\n    if not get_targets:\n        dataset = dataset.map(lambda image, target: image)\n    dataset = dataset.batch(config.BATCH_SIZE)\n    dataset = dataset.prefetch(AUTO)\n    return dataset\n\n# This function is to get our training tensors\ndef get_test_dataset(filenames, get_names = True):\n    dataset = load_dataset(filenames, ordered = True)\n    dataset = dataset.map(data_augment, num_parallel_calls = AUTO)\n    dataset = dataset.map(arcface_inference_format, num_parallel_calls = AUTO)\n    if not get_names:\n        dataset = dataset.map(lambda image, posting_id: image)\n    dataset = dataset.batch(config.BATCH_SIZE)\n    dataset = dataset.prefetch(AUTO)\n    return dataset","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"row = 10; col = 8;\nrow = min(row,config.BATCH_SIZE//col)\nN_TRAIN = count_data_items(train_files)\nprint(N_TRAIN)\nds = get_training_dataset(train_files)\n\nfor (sample,label) in ds:\n    img = sample['inp1']\n    plt.figure(figsize=(25,int(25*row/col)))\n    for j in range(row*col):\n        plt.subplot(row,col,j+1)\n        plt.title(label[j].numpy())\n        plt.axis('off')\n        plt.imshow(img[j,])\n    plt.show()\n    break\nprint(img.shape)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"row = 10; col = 8;\nrow = min(row,config.BATCH_SIZE//col)\nN_TEST = count_data_items(test_files)\nprint(N_TEST)\nds = get_test_dataset(test_files)\n\nfor (img,label) in ds:\n    plt.figure(figsize=(25,int(25*row/col)))\n    for j in range(row*col):\n        plt.subplot(row,col,j+1)\n        plt.title(label[j].numpy())\n        plt.axis('off')\n        plt.imshow(img[j,])\n    plt.show()\n    break\nprint(img.shape)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Model","metadata":{}},{"cell_type":"code","source":"# Arcmarginproduct class keras layer\nclass ArcMarginProduct(tf.keras.layers.Layer):\n    '''\n    Implements large margin arc distance.\n\n    Reference:\n        https://arxiv.org/pdf/1801.07698.pdf\n        https://github.com/lyakaap/Landmark2019-1st-and-3rd-Place-Solution/\n            blob/master/src/modeling/metric_learning.py\n    '''\n    def __init__(self, n_classes, s=30, m=0.50, easy_margin=False,\n                 ls_eps=0.0, **kwargs):\n\n        super(ArcMarginProduct, self).__init__(**kwargs)\n\n        self.n_classes = n_classes\n        self.s = s\n        self.m = m\n        self.ls_eps = ls_eps\n        self.easy_margin = easy_margin\n        self.cos_m = tf.math.cos(m)\n        self.sin_m = tf.math.sin(m)\n        self.th = tf.math.cos(math.pi - m)\n        self.mm = tf.math.sin(math.pi - m) * m\n\n    def get_config(self):\n\n        config = super().get_config().copy()\n        config.update({\n            'n_classes': self.n_classes,\n            's': self.s,\n            'm': self.m,\n            'ls_eps': self.ls_eps,\n            'easy_margin': self.easy_margin,\n        })\n        return config\n\n    def build(self, input_shape):\n        super(ArcMarginProduct, self).build(input_shape[0])\n\n        self.W = self.add_weight(\n            name='W',\n            shape=(int(input_shape[0][-1]), self.n_classes),\n            initializer='glorot_uniform',\n            dtype='float32',\n            trainable=True,\n            regularizer=None)\n\n    def call(self, inputs):\n        X, y = inputs\n        y = tf.cast(y, dtype=tf.int32)\n        cosine = tf.matmul(\n            tf.math.l2_normalize(X, axis=1),\n            tf.math.l2_normalize(self.W, axis=0)\n        )\n        sine = tf.math.sqrt(1.0 - tf.math.pow(cosine, 2))\n        phi = cosine * self.cos_m - sine * self.sin_m\n        if self.easy_margin:\n            phi = tf.where(cosine > 0, phi, cosine)\n        else:\n            phi = tf.where(cosine > self.th, phi, cosine - self.mm)\n        one_hot = tf.cast(\n            tf.one_hot(y, depth=self.n_classes),\n            dtype=cosine.dtype\n        )\n        if self.ls_eps > 0:\n            one_hot = (1 - self.ls_eps) * one_hot + self.ls_eps / self.n_classes\n\n        output = (one_hot * phi) + ((1.0 - one_hot) * cosine)\n        output *= self.s\n        return output","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"EFNS = [efn.EfficientNetB0, efn.EfficientNetB1, efn.EfficientNetB2, efn.EfficientNetB3, \n        efn.EfficientNetB4, efn.EfficientNetB5, efn.EfficientNetB6, efn.EfficientNetB7]\n\ndef freeze_BN(model):\n    # Unfreeze layers while leaving BatchNorm layers frozen\n    for layer in model.layers:\n        if not isinstance(layer, tf.keras.layers.BatchNormalization):\n            layer.trainable = True\n        else:\n            layer.trainable = False\n\n# Function to create our EfficientNetB3 model\ndef get_model():\n\n    if config.head=='arcface':\n        head = ArcMarginProduct\n    else:\n        assert 1==2, \"INVALID HEAD\"\n    \n    with strategy.scope():\n        \n        margin = head(\n            n_classes = config.N_CLASSES, \n            s = 30, \n            m = 0.3, \n            name=f'head/{config.head}', \n            dtype='float32'\n            )\n\n        inp = tf.keras.layers.Input(shape = [config.IMAGE_SIZE, config.IMAGE_SIZE, 3], name = 'inp1')\n        label = tf.keras.layers.Input(shape = (), name = 'inp2')\n        \n        if config.model_type == 'effnetv1':\n            x = EFNS[config.EFF_NET](weights = 'noisy-student', include_top = False)(inp)\n            embed = tf.keras.layers.GlobalAveragePooling2D()(x)\n        elif config.model_type == 'effnetv2':\n            FEATURE_VECTOR = f'{EFFNETV2_ROOT}/tfhub_models/efficientnetv2-{config.EFF_NETV2}/feature_vector'\n            embed = tfhub.KerasLayer(FEATURE_VECTOR, trainable=True)(inp)\n            \n        embed = tf.keras.layers.Dropout(0.2)(embed)\n        embed = tf.keras.layers.Dense(512)(embed)\n        x = margin([embed, label])\n        \n        output = tf.keras.layers.Softmax(dtype='float32')(x)\n        \n        model = tf.keras.models.Model(inputs = [inp, label], outputs = [output])\n        embed_model = tf.keras.models.Model(inputs = inp, outputs = embed)  \n        \n        opt = tf.keras.optimizers.Adam(learning_rate = config.LR)\n        if config.FREEZE_BATCH_NORM:\n            freeze_BN(model)\n\n        model.compile(\n            optimizer = opt,\n            loss    = [ tf.keras.losses.SparseCategoricalCrossentropy()],\n            metrics = [tf.keras.metrics.SparseCategoricalAccuracy(),\n                       tf.keras.metrics.SparseTopKCategoricalAccuracy(k=5)]\n            ) \n\n#         model.compile(\n#               loss=SparseCircleLoss(batch_size=config.BATCH_SIZE),\n#               optimizer=opt,\n#               metrics = [tf.keras.metrics.SparseCategoricalAccuracy(),\n#                          tf.keras.metrics.SparseTopKCategoricalAccuracy(k=5)]\n#                      )\n        \n        return model,embed_model","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def get_lr_callback(plot=False):\n    lr_start   = 0.000001\n    lr_max     = 0.000005 * config.BATCH_SIZE  \n    lr_min     = 0.000001\n    lr_ramp_ep = 4\n    lr_sus_ep  = 0\n    lr_decay   = 0.9\n   \n    def lrfn(epoch):\n        if config.RESUME:\n            epoch = epoch + config.RESUME_EPOCH\n        if epoch < lr_ramp_ep:\n            lr = (lr_max - lr_start) / lr_ramp_ep * epoch + lr_start\n            \n        elif epoch < lr_ramp_ep + lr_sus_ep:\n            lr = lr_max\n            \n        else:\n            lr = (lr_max - lr_min) * lr_decay**(epoch - lr_ramp_ep - lr_sus_ep) + lr_min\n            \n        return lr\n        \n    if plot:\n        epochs = list(range(config.EPOCHS))\n        learning_rates = [lrfn(x) for x in epochs]\n        plt.scatter(epochs,learning_rates)\n        plt.show()\n\n    lr_callback = tf.keras.callbacks.LearningRateScheduler(lrfn, verbose=False)\n    return lr_callback\n\nget_lr_callback(plot=True)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"class Snapshot(tf.keras.callbacks.Callback):\n    \n    def __init__(self,fold,snapshot_epochs=[]):\n        super(Snapshot, self).__init__()\n        self.snapshot_epochs = snapshot_epochs\n        self.fold = fold\n        \n        \n    def on_epoch_end(self, epoch, logs=None):\n        # logs is a dictionary\n#         print(f\"epoch: {epoch}, train_acc: {logs['acc']}, valid_acc: {logs['val_acc']}\")\n        if epoch in self.snapshot_epochs: # your custom condition         \n            self.model.save_weights(config.save_dir+f\"/EF{config.MODEL_NAME}_epoch{epoch}.h5\")\n        self.model.save_weights(config.save_dir+f\"/{config.MODEL_NAME}_last.h5\")","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Train","metadata":{}},{"cell_type":"code","source":"if config.EVALUATE:\n    TRAINING_FILENAMES = [x for i,x in enumerate(train_files) if i%config.FOLDS!=config.FOLD_TO_RUN]\n    VALIDATION_FILENAMES = [x for i,x in enumerate(train_files) if i%config.FOLDS==config.FOLD_TO_RUN]\n    print(len(TRAINING_FILENAMES),len(VALIDATION_FILENAMES),count_data_items(TRAINING_FILENAMES),count_data_items(VALIDATION_FILENAMES))\nelse:\n    TRAINING_FILENAMES = train_files\n    print(len(TRAINING_FILENAMES), count_data_items(TRAINING_FILENAMES))","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"if config.DEBUG:\n    TRAINING_FILENAMES = [TRAINING_FILENAMES[0]]\n    VALIDATION_FILENAMES = [VALIDATION_FILENAMES[0]]\n    print(len(TRAINING_FILENAMES),len(VALIDATION_FILENAMES),count_data_items(TRAINING_FILENAMES),count_data_items(VALIDATION_FILENAMES))\n    test_files = [test_files[0]]","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"seed_everything(config.SEED)\nVERBOSE = 1\ntrain_dataset = get_training_dataset(TRAINING_FILENAMES)\nval_dataset = get_val_dataset(VALIDATION_FILENAMES)\nSTEPS_PER_EPOCH = count_data_items(TRAINING_FILENAMES) // config.BATCH_SIZE\nK.clear_session()\nmodel,embed_model = get_model()\nmodel.summary()","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Evaluation","metadata":{}},{"cell_type":"code","source":"# model.load_weights(\"../input/score-ver-1/effnetv1_b5_loss.h5\")","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"sub = pd.read_csv(\"../input/score-ver-1/submission.csv\")\nsub.to_csv('submission.csv',index=False)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# def get_ids(filename):\n#     ds = get_test_dataset([filename],get_names=True).map(lambda image, image_name: image_name).unbatch()\n#     NUM_IMAGES = count_data_items([filename])\n#     ids = next(iter(ds.batch(NUM_IMAGES))).numpy().astype('U')\n#     return ids\n\n# def get_targets(filename):\n#     ds = get_eval_dataset([filename],get_targets=True).map(lambda image, target: target).unbatch()\n#     NUM_IMAGES = count_data_items([filename])\n#     ids = next(iter(ds.batch(NUM_IMAGES))).numpy()\n#     return ids\n\n# def get_embeddings(filename):\n#     ds = get_test_dataset([filename],get_names=False)\n#     embeddings = embed_model.predict(ds,verbose=0)\n#     return embeddings\n\n# def get_predictions(test_df,threshold=0.2):\n#     predictions = {}\n#     for i,row in tqdm(test_df.iterrows()):\n#         if row.image in predictions:\n#             if len(predictions[row.image])==5:\n#                 continue\n#             predictions[row.image].append(row.target)\n#         elif row.confidence>threshold:\n#             predictions[row.image] = [row.target,'new_individual']\n#         else:\n#             predictions[row.image] = ['new_individual',row.target]\n\n#     for x in tqdm(predictions):\n#         if len(predictions[x])<5:\n#             remaining = [y for y in sample_list if y not in predictions]\n#             predictions[x] = predictions[x]+remaining\n#             predictions[x] = predictions[x][:5]\n        \n#     return predictions\n\n# def map_per_image(label, predictions):\n#     \"\"\"Computes the precision score of one image.\n\n#     Parameters\n#     ----------\n#     label : string\n#             The true label of the image\n#     predictions : list\n#             A list of predicted elements (order does matter, 5 predictions allowed per image)\n\n#     Returns\n#     -------\n#     score : double\n#     \"\"\"    \n#     try:\n#         return 1 / (predictions[:5].index(label) + 1)\n#     except ValueError:\n#         return 0.0\n    \n# f = open ('../input/happywhalelabelidx/individual_ids.json', \"r\")\n# target_encodings = json.loads(f.read())\n# target_encodings = {target_encodings[x]:x for x in target_encodings}\n# sample_list = ['938b7e931166', '5bf17305f073', '7593d2aee842', '7362d7a01d00','956562ff2888']","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# train_targets = []\n# train_embeddings = []\n# for filename in tqdm(TRAINING_FILENAMES):\n#     embeddings = get_embeddings(filename)\n#     targets = get_targets(filename)\n#     train_embeddings.append(embeddings)\n#     train_targets.append(targets)\n# train_embeddings = np.concatenate(train_embeddings)\n# train_targets = np.concatenate(train_targets)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# train_embeddings","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# def interpolate_nans(X):\n#     \"\"\"Overwrite NaNs with column value interpolations.\"\"\"\n#     for j in range(X.shape[1]):\n#         mask_j = np.isnan(X[:,j])\n#         X[mask_j,j] = np.interp(np.flatnonzero(mask_j), np.flatnonzero(~mask_j), X[~mask_j,j])\n#     return X","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# train_embeddings = interpolate_nans(train_embeddings)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# from sklearn.neighbors import NearestNeighbors\n# neigh = NearestNeighbors(n_neighbors=config.KNN,metric='cosine')\n# neigh.fit(train_embeddings)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# test_ids = []\n# test_nn_distances = []\n# test_nn_idxs = []\n# val_targets = []\n# val_embeddings = []\n# for filename in tqdm(VALIDATION_FILENAMES[:]):\n#     embeddings = get_embeddings(filename)\n#     targets = get_targets(filename)\n#     ids = get_ids(filename)\n\n#     targets = targets[~np.isnan(embeddings).any(axis = 1)]\n#     ids = ids[~np.isnan(embeddings).any(axis = 1)]\n#     embeddings = embeddings[~np.isnan(embeddings).any(axis = 1)]\n#     distances,idxs = neigh.kneighbors(embeddings, config.KNN, return_distance=True)\n#     test_ids.append(ids)\n#     test_nn_idxs.append(idxs)\n#     test_nn_distances.append(distances)\n#     val_embeddings.append(embeddings)\n#     val_targets.append(targets)\n# test_nn_distances = np.concatenate(test_nn_distances)\n# test_nn_idxs = np.concatenate(test_nn_idxs)\n# test_ids = np.concatenate(test_ids)\n# val_embeddings = np.concatenate(val_embeddings)\n# val_targets = np.concatenate(val_targets)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# allowed_targets = set([target_encodings[x] for x in np.unique(train_targets)])\n# val_targets_df = pd.DataFrame(np.stack([test_ids,val_targets],axis=1),columns=['image','target'])\n# val_targets_df['target'] = val_targets_df['target'].astype(int).map(target_encodings)\n# val_targets_df.loc[~val_targets_df.target.isin(allowed_targets),'target'] = 'new_individual'\n# val_targets_df.target.value_counts()","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# test_df = []\n# for i in tqdm(range(len(test_ids))):\n#     id_ = test_ids[i]\n#     targets = train_targets[test_nn_idxs[i]]\n#     distances = test_nn_distances[i]\n#     subset_preds = pd.DataFrame(np.stack([targets,distances],axis=1),columns=['target','distances'])\n#     subset_preds['image'] = id_\n#     test_df.append(subset_preds)\n# test_df = pd.concat(test_df).reset_index(drop=True)\n# test_df['confidence'] = 1-test_df['distances']\n# test_df = test_df.groupby(['image','target']).confidence.max().reset_index()\n# test_df = test_df.sort_values('confidence',ascending=False).reset_index(drop=True)\n# test_df['target'] = test_df['target'].map(target_encodings)\n# test_df.to_csv('val_neighbors.csv')\n# test_df.image.value_counts().value_counts()","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# best_th = 0\n# best_cv = 0\n# for th in [0.1*x for x in range(11)]:\n#     all_preds = get_predictions(test_df,threshold=th)\n#     cv = 0\n#     for i,row in val_targets_df.iterrows():\n#         target = row.target\n#         preds = all_preds[row.image]\n#         val_targets_df.loc[i,th] = map_per_image(target,preds)\n#     cv = val_targets_df[th].mean()\n#     print(f\"CV at threshold {th}: {cv}\")\n#     if cv>best_cv:\n#         best_th = th\n#         best_cv = cv\n","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# print(\"Best threshold\",best_th)\n# print(\"Best cv\",best_cv)\n# val_targets_df.describe()","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# ## Adjustment: Since Public lb has nearly 10% 'new_individual' (Be Careful for private LB)\n# val_targets_df['is_new_individual'] = val_targets_df.target=='new_individual'\n# print(val_targets_df.is_new_individual.value_counts().to_dict())\n# val_scores = val_targets_df.groupby('is_new_individual').mean().T\n# val_scores['adjusted_cv'] = val_scores[True]*0.1+val_scores[False]*0.9\n# best_threshold_adjusted = val_scores['adjusted_cv'].idxmax()\n# print(\"best_threshold\",best_threshold_adjusted)\n# val_scores\n","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Inference","metadata":{}},{"cell_type":"code","source":"# train_embeddings = np.concatenate([train_embeddings,val_embeddings])\n# train_targets = np.concatenate([train_targets,val_targets])\n# print(train_embeddings.shape,train_targets.shape)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# from sklearn.neighbors import NearestNeighbors\n# neigh = NearestNeighbors(n_neighbors=config.KNN,metric='cosine')\n# neigh.fit(train_embeddings)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# test_ids = []\n# test_nn_distances = []\n# test_nn_idxs = []\n# for filename in tqdm(test_files):\n#     embeddings = get_embeddings(filename)\n#     ids = get_ids(filename)\n#     distances,idxs = neigh.kneighbors(embeddings, config.KNN, return_distance=True)\n#     test_ids.append(ids)\n#     test_nn_idxs.append(idxs)\n#     test_nn_distances.append(distances)\n# test_nn_distances = np.concatenate(test_nn_distances)\n# test_nn_idxs = np.concatenate(test_nn_idxs)\n# test_ids = np.concatenate(test_ids)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# sample_submission = pd.read_csv('../input/happy-whale-and-dolphin/sample_submission.csv',index_col='image')\n# print(len(sample_submission))\n# test_df = []\n# for i in tqdm(range(len(test_ids))):\n#     id_ = test_ids[i]\n#     targets = train_targets[test_nn_idxs[i]]\n#     distances = test_nn_distances[i]\n#     subset_preds = pd.DataFrame(np.stack([targets,distances],axis=1),columns=['target','distances'])\n#     subset_preds['image'] = id_\n#     test_df.append(subset_preds)\n# test_df = pd.concat(test_df).reset_index(drop=True)\n# test_df['confidence'] = 1-test_df['distances']\n# test_df = test_df.groupby(['image','target']).confidence.max().reset_index()\n# test_df = test_df.sort_values('confidence',ascending=False).reset_index(drop=True)\n# test_df['target'] = test_df['target'].map(target_encodings)\n# test_df.to_csv('test_neighbors.csv')\n# test_df.image.value_counts().value_counts()","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# len(test_df)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# sample_list = ['938b7e931166', '5bf17305f073', '7593d2aee842', '7362d7a01d00','956562ff2888']","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# predictions = {}\n# for i,row in tqdm(test_df.iterrows()):\n#     if row.image in predictions:\n#         if len(predictions[row.image])==5:\n#             continue\n#         predictions[row.image].append(row.target)\n#     elif row.confidence>best_threshold_adjusted:\n#         predictions[row.image] = [row.target,'new_individual']\n#     else:\n#         predictions[row.image] = ['new_individual',row.target]\n        \n# for x in tqdm(predictions):\n#     if len(predictions[x])<5:\n#         remaining = [y for y in sample_list if y not in predictions]\n#         predictions[x] = predictions[x]+remaining\n#         predictions[x] = predictions[x][:5]\n#     predictions[x] = ' '.join(predictions[x])\n    \n# predictions = pd.Series(predictions).reset_index()\n# predictions.columns = ['image','predictions']\n","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# len(predictions)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# drop the images which do not have a backfin & fill these predictions with a general model such as from: \n\nhttps://www.kaggle.com/aikhmelnytskyy/happywhale-arcface-baseline-eff-net-kfold5-0-652","metadata":{}},{"cell_type":"code","source":"# df2 = pd.read_csv(\"../input/score-ver-1/submission.csv\")\n# ids = np.load(\"../input/score-ver-1/ids_without_backfin.npy\", allow_pickle = True) #whales without backfins determined with a simple classifier","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# ids2 = df2[\"image\"][~df2[\"image\"].isin(predictions[\"image\"])] #images without a bounding box","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# submission = pd.concat([\n#     predictions[~(predictions[\"image\"].isin(ids))],\n#     df2[df2[\"image\"].isin(ids)],\n#     df2[df2[\"image\"].isin(ids2)]\n# ])","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# submission = submission.drop_duplicates()","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# submission.to_csv('submission.csv',index=False)\n# submission.head()","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# len(submission)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]}]}