{"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":"# [Reference](https://keras.io/examples/vision/mobilevit/)","metadata":{}},{"cell_type":"markdown","source":"# Introduction\n\nIn this example, we implement the MobileViT architecture (Mehta et al.), which combines the benefits of Transformers (Vaswani et al.) and convolutions. With Transformers, we can capture long-range dependencies that result in global representations. With convolutions, we can capture spatial relationships that model locality.\n\nBesides combining the properties of Transformers and convolutions, the authors introduce MobileViT as a general-purpose mobile-friendly backbone for different image recognition tasks. Their findings suggest that, performance-wise, MobileViT is better than other models with the same or higher complexity (MobileNetV3, for example), while being efficient on mobile devices.","metadata":{}},{"cell_type":"code","source":"import numpy as np \nimport pandas as pd \nimport matplotlib.pyplot as plt \nfrom PIL import Image\nimport os, cv2, gc\nos.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.preprocessing import LabelEncoder\nimport tensorflow as tf \nfrom tensorflow import keras \nfrom tensorflow.keras import callbacks\nfrom tensorflow.keras import layers, losses, optimizers, metrics\nimport tensorflow_hub as hub\nfrom keras.applications import imagenet_utils\n\n#import tensorflow_datasets as tfds\nimport tensorflow_addons as tfa\nfrom keras.layers.advanced_activations import LeakyReLU\n\ntry:\n    physical_devices = tf.config.list_physical_devices('GPU')\n    tf.config.experimental.set_memory_growth(physical_devices[0], True)\nexcept:\n    pass \n\n# enable mixed_precision and jit compiler \ntf.keras.mixed_precision.set_global_policy('mixed_float16')\ntf.config.optimizer.set_jit(True)","metadata":{"execution":{"iopub.status.busy":"2022-03-17T09:07:09.410483Z","iopub.execute_input":"2022-03-17T09:07:09.410745Z","iopub.status.idle":"2022-03-17T09:07:13.503386Z","shell.execute_reply.started":"2022-03-17T09:07:09.410678Z","shell.execute_reply":"2022-03-17T09:07:13.502614Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"INP_SIZE      = (512, 512) # Input size of the Image Resizer Module (IRM)\nTARGET_SIZE   = (224, 224) # Output size of IRM and Input size of the Vision Transformer \nINTERPOLATION = \"bilinear\"\nN_CLASSES = 15587\n#config_head = 'arcface'\n\n\nNUM_FOLDS  = 5\nBATCH_SIZE = 24\nSEED       = 42\n\nDATA_DIR  = '../input/happy-whale-and-dolphin/'\nTRAIN_DIR = DATA_DIR + 'train_images/'\nTEST_DIR  = DATA_DIR + 'test_images/'\n\n# SetAutoTune\nAUTOTUNE = tf.data.AUTOTUNE  ","metadata":{"execution":{"iopub.status.busy":"2022-03-17T09:07:13.505527Z","iopub.execute_input":"2022-03-17T09:07:13.505817Z","iopub.status.idle":"2022-03-17T09:07:13.512389Z","shell.execute_reply.started":"2022-03-17T09:07:13.505782Z","shell.execute_reply":"2022-03-17T09:07:13.511717Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Prepare tf.data object","metadata":{}},{"cell_type":"code","source":"#https://www.kaggle.com/ipythonx/tf-keras-learning-to-resize-image-for-vit-model/notebook\ndef build_augmenter(is_labelled):\n    def augment(img):\n        img = tf.image.random_flip_left_right(img)\n        img = tf.image.random_flip_up_down(img)\n        img = tf.image.random_saturation(img, 0.65, 1.05)\n        img = tf.image.random_brightness(img, 0.05)\n        img = tf.image.random_contrast(img, 0.75, 1.05)\n        img = tf.image.random_hue(img, 0.05)\n        return img\n    \n    def augment_with_labels(img, label):\n        return augment(img), label\n    return augment_with_labels if is_labelled else augment\n\ndef build_decoder(is_labelled, size):\n    def decode(path):\n        file_bytes = tf.io.read_file(path)\n        img = tf.image.decode_jpeg(file_bytes, channels = 3)\n        img = tf.image.resize(img, (size[0], size[1]))\n        return tf.cast(tf.divide(img, 255.),tf.float32)\n    \n    def decode_with_labels(path, label):\n        label = tf.cast(label, tf.int32)\n        return decode(path),label\n    \n    return decode_with_labels if is_labelled else decode\n\ndef create_dataset(df, \n                   batch_size  = 32, \n                   is_labelled = False, \n                   augment     = False, \n                   repeat      = False, \n                   shuffle     = False,\n                   size        = INP_SIZE):\n    decode_fn    = build_decoder(is_labelled, size)\n    augmenter_fn = build_augmenter(is_labelled)\n    \n    # Create Dataset\n    if is_labelled:\n        dataset = tf.data.Dataset.from_tensor_slices((df['Id'].values, df['target_value'].values))\n    else:\n        dataset = tf.data.Dataset.from_tensor_slices((df['Id'].values))\n        \n    dataset = dataset.map(decode_fn, num_parallel_calls = AUTOTUNE)\n    dataset = dataset.map(augmenter_fn, num_parallel_calls = AUTOTUNE) if augment else dataset\n    dataset = dataset.repeat() if repeat else dataset\n    dataset = dataset.shuffle(1024, reshuffle_each_iteration = True) if shuffle else dataset\n    dataset = dataset.batch(batch_size)\n    dataset = dataset.prefetch(AUTOTUNE)\n    return dataset","metadata":{"execution":{"iopub.status.busy":"2022-03-17T09:07:13.513738Z","iopub.execute_input":"2022-03-17T09:07:13.514318Z","iopub.status.idle":"2022-03-17T09:07:13.528971Z","shell.execute_reply.started":"2022-03-17T09:07:13.514281Z","shell.execute_reply":"2022-03-17T09:07:13.528200Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"label_encoder = LabelEncoder()\n","metadata":{"execution":{"iopub.status.busy":"2022-03-17T09:07:13.532462Z","iopub.execute_input":"2022-03-17T09:07:13.532873Z","iopub.status.idle":"2022-03-17T09:07:13.540266Z","shell.execute_reply.started":"2022-03-17T09:07:13.532844Z","shell.execute_reply":"2022-03-17T09:07:13.539401Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Load Train Data\ntrain_df = pd.read_csv(f'{DATA_DIR}train.csv')\ntrain_df['Id'] = train_df['image'].apply(lambda x: f'{TRAIN_DIR}{x}')\n\n# Adjust typos in \"species\" column from Andrada's kernel\ntrain_df[\"species\"] = train_df[\"species\"].replace([\"bottlenose_dolpin\", \"kiler_whale\",\n                                             \"beluga\", \n                                             \"globis\", \"pilot_whale\"],\n                                            [\"bottlenose_dolphin\", \"killer_whale\",\n                                             \"beluga_whale\", \n                                             \"short_finned_pilot_whale\", \"short_finned_pilot_whale\"])\n\n\n# Set a specific label to be able to perform stratification\n#train_df['stratify_label'] = train_df['individual_id']\n\ntrain_df['target_value']  = label_encoder.fit_transform(train_df['individual_id'] )\n\n# Summary\nprint(f'train_df: {train_df.shape}')\ntrain_df.head()","metadata":{"execution":{"iopub.status.busy":"2022-03-17T09:07:13.542644Z","iopub.execute_input":"2022-03-17T09:07:13.543268Z","iopub.status.idle":"2022-03-17T09:07:13.713068Z","shell.execute_reply.started":"2022-03-17T09:07:13.543230Z","shell.execute_reply":"2022-03-17T09:07:13.712390Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_df[train_df['individual_id']=='19fbb960f07d']","metadata":{"execution":{"iopub.status.busy":"2022-03-17T10:44:00.341300Z","iopub.execute_input":"2022-03-17T10:44:00.341686Z","iopub.status.idle":"2022-03-17T10:44:00.365251Z","shell.execute_reply.started":"2022-03-17T10:44:00.341645Z","shell.execute_reply":"2022-03-17T10:44:00.364457Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_df.species.value_counts()","metadata":{"execution":{"iopub.status.busy":"2022-03-17T10:42:12.903048Z","iopub.execute_input":"2022-03-17T10:42:12.903388Z","iopub.status.idle":"2022-03-17T10:42:12.924220Z","shell.execute_reply.started":"2022-03-17T10:42:12.903349Z","shell.execute_reply":"2022-03-17T10:42:12.923342Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_df.isnull().sum()","metadata":{"execution":{"iopub.status.busy":"2022-03-17T09:07:13.714497Z","iopub.execute_input":"2022-03-17T09:07:13.714990Z","iopub.status.idle":"2022-03-17T09:07:13.743074Z","shell.execute_reply.started":"2022-03-17T09:07:13.714952Z","shell.execute_reply":"2022-03-17T09:07:13.742340Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_ds, val_ds = train_df[:30000], train_df[30000:]","metadata":{"execution":{"iopub.status.busy":"2022-03-17T09:07:13.745933Z","iopub.execute_input":"2022-03-17T09:07:13.746177Z","iopub.status.idle":"2022-03-17T09:07:13.752035Z","shell.execute_reply.started":"2022-03-17T09:07:13.746153Z","shell.execute_reply":"2022-03-17T09:07:13.751288Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Load Test Data\ntest_df = pd.read_csv(f'{DATA_DIR}sample_submission.csv')\ntest_df['Id'] = test_df['image'].apply(lambda x: f'{TEST_DIR}{x}')\ntest_df['individual_id'] = 0\n\n# Summary\nprint(f'test_df: {test_df.shape}')\ntest_df.head()","metadata":{"execution":{"iopub.status.busy":"2022-03-17T09:07:13.753336Z","iopub.execute_input":"2022-03-17T09:07:13.753801Z","iopub.status.idle":"2022-03-17T09:07:13.830959Z","shell.execute_reply.started":"2022-03-17T09:07:13.753757Z","shell.execute_reply":"2022-03-17T09:07:13.830251Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# MobileViT","metadata":{}},{"cell_type":"markdown","source":"MobileViT architecture comprised of following blocks\n* Strided 3x3 convolutions that process the input image\n* MobileNetV2-style inverted residual blocks for downsampling the resolution of the intermediate feature maps\n* MobileViT blocks that combine the benefits of Transformers and convolutions. It is presented in the figure below (taken from the original paper):","metadata":{}},{"cell_type":"markdown","source":"![Screenshot from 2022-03-16 22-52-00.png](attachment:90baadc6-81f6-440f-8136-b0289236341f.png)","metadata":{},"attachments":{"90baadc6-81f6-440f-8136-b0289236341f.png":{"image/png":"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"}}},{"cell_type":"code","source":"# Values are from table 4.\npatch_size = 4  # 2x2, for the Transformer blocks.\nimage_size = 256\nexpansion_factor = 2  # expansion factor for the MobileNetV2 blocks.","metadata":{"execution":{"iopub.status.busy":"2022-03-17T09:07:13.832319Z","iopub.execute_input":"2022-03-17T09:07:13.832776Z","iopub.status.idle":"2022-03-17T09:07:13.836608Z","shell.execute_reply.started":"2022-03-17T09:07:13.832738Z","shell.execute_reply":"2022-03-17T09:07:13.835832Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def conv_block(x, filters=16, kernel_size=3, strides=2):\n    conv_layer = layers.Conv2D(\n        filters, kernel_size, strides=strides, activation=tf.nn.swish, padding=\"same\"\n    )\n    return conv_layer(x)\n\n\n# Reference: https://git.io/JKgtC\n\n\ndef inverted_residual_block(x, expanded_channels, output_channels, strides=1):\n    m = layers.Conv2D(expanded_channels, 1, padding=\"same\", use_bias=False)(x)\n    m = layers.BatchNormalization()(m)\n    m = tf.nn.swish(m)\n\n    if strides == 2:\n        m = layers.ZeroPadding2D(padding=imagenet_utils.correct_pad(m, 3))(m)\n    m = layers.DepthwiseConv2D(\n        3, strides=strides, padding=\"same\" if strides == 1 else \"valid\", use_bias=False\n    )(m)\n    m = layers.BatchNormalization()(m)\n    m = tf.nn.swish(m)\n\n    m = layers.Conv2D(output_channels, 1, padding=\"same\", use_bias=False)(m)\n    m = layers.BatchNormalization()(m)\n\n    if tf.math.equal(x.shape[-1], output_channels) and strides == 1:\n        return layers.Add()([m, x])\n    return m\n\n\n# Reference:\n# https://keras.io/examples/vision/image_classification_with_vision_transformer/\n\n\ndef mlp(x, hidden_units, dropout_rate):\n    for units in hidden_units:\n        x = layers.Dense(units, activation=tf.nn.swish)(x)\n        x = layers.Dropout(dropout_rate)(x)\n    return x\n\n\ndef transformer_block(x, transformer_layers, projection_dim, num_heads=2):\n    for _ in range(transformer_layers):\n        # Layer normalization 1.\n        x1 = layers.LayerNormalization(epsilon=1e-6)(x)\n        # Create a multi-head attention layer.\n        attention_output = layers.MultiHeadAttention(\n            num_heads=num_heads, key_dim=projection_dim, dropout=0.1\n        )(x1, x1)\n        # Skip connection 1.\n        x2 = layers.Add()([attention_output, x])\n        # Layer normalization 2.\n        x3 = layers.LayerNormalization(epsilon=1e-6)(x2)\n        # MLP.\n        x3 = mlp(x3, hidden_units=[x.shape[-1] * 2, x.shape[-1]], dropout_rate=0.1,)\n        # Skip connection 2.\n        x = layers.Add()([x3, x2])\n\n    return x\n\n\ndef mobilevit_block(x, num_blocks, projection_dim, strides=1):\n    # Local projection with convolutions.\n    local_features = conv_block(x, filters=projection_dim, strides=strides)\n    local_features = conv_block(\n        local_features, filters=projection_dim, kernel_size=1, strides=strides\n    )\n\n    # Unfold into patches and then pass through Transformers.\n    num_patches = int((local_features.shape[1] * local_features.shape[2]) / patch_size)\n    non_overlapping_patches = layers.Reshape((patch_size, num_patches, projection_dim))(\n        local_features\n    )\n    global_features = transformer_block(\n        non_overlapping_patches, num_blocks, projection_dim\n    )\n\n    # Fold into conv-like feature-maps.\n    folded_feature_map = layers.Reshape((*local_features.shape[1:-1], projection_dim))(\n        global_features\n    )\n\n    # Apply point-wise conv -> concatenate with the input features.\n    folded_feature_map = conv_block(\n        folded_feature_map, filters=x.shape[-1], kernel_size=1, strides=strides\n    )\n    local_global_features = layers.Concatenate(axis=-1)([x, folded_feature_map])\n\n    # Fuse the local and global features using a convoluion layer.\n    local_global_features = conv_block(\n        local_global_features, filters=projection_dim, strides=strides\n    )\n\n    return local_global_features\n","metadata":{"execution":{"iopub.status.busy":"2022-03-17T09:07:13.838173Z","iopub.execute_input":"2022-03-17T09:07:13.838761Z","iopub.status.idle":"2022-03-17T09:07:13.859257Z","shell.execute_reply.started":"2022-03-17T09:07:13.838709Z","shell.execute_reply":"2022-03-17T09:07:13.858481Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# More on the MobileViT block:\n\n*     First, the feature representations (A) go through convolution blocks that capture local relationships. The expected shape of a single entry here would be (h, w, num_channels).\n*     Then they get unfolded into another vector with shape (p, n, num_channels), where p is the area of a small patch, and n is (h * w) / p. So, we end up with n non-overlapping patches.\n*     This unfolded vector is then passed through a Tranformer block that captures global relationships between the patches.\n*     The output vector (B) is again folded into a vector of shape (h, w, num_channels) resembling a feature map coming out of convolutions.\n\nVectors A and B are then passed through two more convolutional layers to fuse the local and global representations. Notice how the spatial resolution of the final vector remains unchanged at this point. The authors also present an explanation of how the MobileViT block resembles a convolution block of a CNN. For more details, please refer to the original paper.\n\nNext, we combine these blocks together and implement the MobileViT architecture (XXS variant). The following figure (taken from the original paper) presents a schematic representation of the architecture:","metadata":{}},{"cell_type":"markdown","source":"![Screenshot from 2022-03-16 22-55-19.png](attachment:7e6962e8-de9f-425b-b4a8-4507d44827d1.png)","metadata":{},"attachments":{"7e6962e8-de9f-425b-b4a8-4507d44827d1.png":{"image/png":"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"}}},{"cell_type":"code","source":"def create_mobilevit(num_classes=N_CLASSES):\n    inputs = keras.Input((image_size, image_size, 3))\n    \n    x = layers.Rescaling(scale=1.0 / 255)(inputs)\n\n    # Initial conv-stem -> MV2 block.\n    x = conv_block(x, filters=16)\n    x = inverted_residual_block(\n        x, expanded_channels=16 * expansion_factor, output_channels=16\n    )\n\n    # Downsampling with MV2 block.\n    x = inverted_residual_block(\n        x, expanded_channels=16 * expansion_factor, output_channels=24, strides=2\n    )\n    x = inverted_residual_block(\n        x, expanded_channels=24 * expansion_factor, output_channels=24\n    )\n    x = inverted_residual_block(\n        x, expanded_channels=24 * expansion_factor, output_channels=24\n    )\n\n    # First MV2 -> MobileViT block.\n    x = inverted_residual_block(\n        x, expanded_channels=24 * expansion_factor, output_channels=48, strides=2\n    )\n    x = mobilevit_block(x, num_blocks=2, projection_dim=64)\n\n    # Second MV2 -> MobileViT block.\n    x = inverted_residual_block(\n        x, expanded_channels=64 * expansion_factor, output_channels=64, strides=2\n    )\n    x = mobilevit_block(x, num_blocks=4, projection_dim=80)\n\n    # Third MV2 -> MobileViT block.\n    x = inverted_residual_block(\n        x, expanded_channels=80 * expansion_factor, output_channels=80, strides=2\n    )\n    x = mobilevit_block(x, num_blocks=3, projection_dim=96)\n    x = conv_block(x, filters=320, kernel_size=1, strides=1)\n\n    # Classification head.\n    x = layers.GlobalAvgPool2D()(x)\n    x = layers.Dropout(0.3)(x)\n    outputs = layers.Dense(num_classes,kernel_regularizer='l2', activation=\"softmax\")(x)\n\n    return keras.Model(inputs, outputs)\n\n\nmobilevit_xxs = create_mobilevit()\n#mobilevit_xxs.summary()","metadata":{"execution":{"iopub.status.busy":"2022-03-17T09:07:13.861675Z","iopub.execute_input":"2022-03-17T09:07:13.862178Z","iopub.status.idle":"2022-03-17T09:07:16.103322Z","shell.execute_reply.started":"2022-03-17T09:07:13.862144Z","shell.execute_reply":"2022-03-17T09:07:16.102607Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"mobilevit_xxs.summary()","metadata":{"_kg_hide-output":true,"execution":{"iopub.status.busy":"2022-03-17T09:07:16.104708Z","iopub.execute_input":"2022-03-17T09:07:16.104947Z","iopub.status.idle":"2022-03-17T09:07:16.191901Z","shell.execute_reply.started":"2022-03-17T09:07:16.104913Z","shell.execute_reply":"2022-03-17T09:07:16.191200Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"batch_size = 128\nauto = tf.data.AUTOTUNE\nresize_bigger = 280","metadata":{"execution":{"iopub.status.busy":"2022-03-17T09:07:16.194983Z","iopub.execute_input":"2022-03-17T09:07:16.195179Z","iopub.status.idle":"2022-03-17T09:07:16.198620Z","shell.execute_reply.started":"2022-03-17T09:07:16.195154Z","shell.execute_reply":"2022-03-17T09:07:16.197873Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"individual_id_group = train_df['target_value'].copy().to_list()","metadata":{"execution":{"iopub.status.busy":"2022-03-17T09:07:16.200143Z","iopub.execute_input":"2022-03-17T09:07:16.200428Z","iopub.status.idle":"2022-03-17T09:07:16.210255Z","shell.execute_reply.started":"2022-03-17T09:07:16.200343Z","shell.execute_reply":"2022-03-17T09:07:16.209586Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Training , Valdation Dataset ","metadata":{}},{"cell_type":"code","source":"trn_ds, val_ds = train_test_split(train_df,test_size=0.33, random_state=42) ","metadata":{"execution":{"iopub.status.busy":"2022-03-17T09:07:16.211817Z","iopub.execute_input":"2022-03-17T09:07:16.212082Z","iopub.status.idle":"2022-03-17T09:07:16.233401Z","shell.execute_reply.started":"2022-03-17T09:07:16.212048Z","shell.execute_reply":"2022-03-17T09:07:16.232742Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#train_df =  train_df.sample(frac = 0.1)\ntrn_ds.shape, val_ds.shape","metadata":{"execution":{"iopub.status.busy":"2022-03-17T09:07:16.234708Z","iopub.execute_input":"2022-03-17T09:07:16.235164Z","iopub.status.idle":"2022-03-17T09:07:16.241858Z","shell.execute_reply.started":"2022-03-17T09:07:16.235128Z","shell.execute_reply":"2022-03-17T09:07:16.240406Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"training_dataset = create_dataset(trn_ds,\n                                  batch_size  = BATCH_SIZE, \n                                  is_labelled = True, \n                                  augment = True,\n                                  repeat  = False, \n                                  shuffle = False,\n                                 size = (image_size, image_size))\n#sample_train_images, _ = next(iter(training_dataset))","metadata":{"execution":{"iopub.status.busy":"2022-03-17T09:07:16.243666Z","iopub.execute_input":"2022-03-17T09:07:16.245698Z","iopub.status.idle":"2022-03-17T09:07:16.501789Z","shell.execute_reply.started":"2022-03-17T09:07:16.245665Z","shell.execute_reply":"2022-03-17T09:07:16.501036Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"batch, labels = next(iter(training_dataset))","metadata":{"execution":{"iopub.status.busy":"2022-03-17T09:07:16.503136Z","iopub.execute_input":"2022-03-17T09:07:16.503370Z","iopub.status.idle":"2022-03-17T09:07:17.427598Z","shell.execute_reply.started":"2022-03-17T09:07:16.503335Z","shell.execute_reply":"2022-03-17T09:07:17.426811Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"plt.figure(figsize=(16, 10))\nfor i, image in enumerate(batch[:20]):\n    ax = plt.subplot(5, 4, i + 1)\n    plt.title(labels[i].numpy())\n    plt.imshow(image.numpy().squeeze())\n    plt.axis(\"off\")\n\n\nplt.tight_layout()","metadata":{"execution":{"iopub.status.busy":"2022-03-17T09:07:17.429553Z","iopub.execute_input":"2022-03-17T09:07:17.429824Z","iopub.status.idle":"2022-03-17T09:07:18.692179Z","shell.execute_reply.started":"2022-03-17T09:07:17.429790Z","shell.execute_reply":"2022-03-17T09:07:18.691537Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"val_dataset = create_dataset(val_ds,\n                                  batch_size  = BATCH_SIZE, \n                                  is_labelled = True, \n                                  augment = True,\n                                  repeat  = False, \n                                  shuffle = False,\n                            size = (image_size, image_size))\n#sample_val_images, _ = next(iter(val_dataset))","metadata":{"execution":{"iopub.status.busy":"2022-03-17T09:07:18.693106Z","iopub.execute_input":"2022-03-17T09:07:18.693327Z","iopub.status.idle":"2022-03-17T09:07:18.759142Z","shell.execute_reply.started":"2022-03-17T09:07:18.693297Z","shell.execute_reply":"2022-03-17T09:07:18.758494Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"batch, labels = next(iter(val_dataset))","metadata":{"execution":{"iopub.status.busy":"2022-03-17T09:07:18.760486Z","iopub.execute_input":"2022-03-17T09:07:18.760766Z","iopub.status.idle":"2022-03-17T09:07:19.557275Z","shell.execute_reply.started":"2022-03-17T09:07:18.760729Z","shell.execute_reply":"2022-03-17T09:07:19.556477Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"plt.figure(figsize=(16, 10))\nfor i, image in enumerate(batch[:20]):\n    ax = plt.subplot(5, 4,  i + 1)\n    plt.title(labels[i].numpy())\n    plt.imshow(image.numpy().squeeze())\n    plt.axis(\"off\")\n\n\nplt.tight_layout()","metadata":{"execution":{"iopub.status.busy":"2022-03-17T09:07:19.558996Z","iopub.execute_input":"2022-03-17T09:07:19.559248Z","iopub.status.idle":"2022-03-17T09:07:21.239070Z","shell.execute_reply.started":"2022-03-17T09:07:19.559212Z","shell.execute_reply":"2022-03-17T09:07:21.238436Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Train a MobileViT (XXS) model","metadata":{}},{"cell_type":"code","source":"#unq_lbls = train_df.target_value.unique()\n\nfrom sklearn.utils import class_weight\nclass_wgt = dict(zip(np.unique(train_df.target_value), class_weight.compute_class_weight(class_weight = 'balanced', classes = np.unique(train_df.target_value), \n                y = train_df.target_value))) \n#class_wgt = class_weight.compute_class_weight(class_weight = 'balanced',classes = np.unique(train_df.target_value),y = train_df.target_value )","metadata":{"execution":{"iopub.status.busy":"2022-03-17T11:45:42.841935Z","iopub.execute_input":"2022-03-17T11:45:42.842571Z","iopub.status.idle":"2022-03-17T11:45:42.882255Z","shell.execute_reply.started":"2022-03-17T11:45:42.842509Z","shell.execute_reply":"2022-03-17T11:45:42.881537Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"learning_rate = 0.002\nlabel_smoothing_factor = 0.1\nepochs = 3\n\n#optimizer = tf.keras.optimizers.RMSprop(learning_rate=learning_rate,decay=1e-6, momentum=0.9, clipnorm=1. )\noptimizer = tf.keras.optimizers.Adam(learning_rate=learning_rate)\n#loss_fn = keras.losses.CategoricalCrossentropy(label_smoothing=label_smoothing_factor)\nloss_fn = keras.losses.SparseCategoricalCrossentropy()\n\n\ndef run_experiment(model, fold, epochs=epochs):\n    tf.keras.backend.clear_session()\n    gc.collect()\n    if fold < 1 :\n        \n        model.compile(optimizer=optimizer, \n                              loss=loss_fn,\n                              #metrics = ['accuracy'])\n                              metrics = [tf.keras.metrics.SparseCategoricalAccuracy(),tf.keras.metrics.SparseTopKCategoricalAccuracy(k=5)])\n\n        checkpoint_filepath = \"/tmp/checkpoint\"\n        checkpoint_callback = keras.callbacks.ModelCheckpoint(\n            checkpoint_filepath,\n            monitor=\"val_sparse_categorical_accuracy\",\n            save_best_only=True,\n            save_weights_only=True,\n        )\n\n\n        trn = train_df.iloc[train_index]\n        val = train_df.iloc[val_index]\n        training_dataset = create_dataset(trn, \n                                          batch_size  = BATCH_SIZE, \n                                          is_labelled = True, \n                                          augment     = True, \n                                          repeat      = False, \n                                          shuffle     = False,\n                                         size = (image_size, image_size))\n        val_dataset = create_dataset(val, \n                                            batch_size  = BATCH_SIZE, \n                                            is_labelled = True,\n                                            augment     = True, \n                                            repeat      = False,\n                                            shuffle     = False,\n                                           size = (image_size, image_size))\n\n\n        history = model.fit(\n            training_dataset,\n            validation_data=val_dataset,\n            epochs=epochs,\n            class_weight = class_wgt,\n            callbacks=[checkpoint_callback],\n        )\n        #model.summary()\n        model.load_weights(checkpoint_filepath)\n        _,_, accuracy = model.evaluate(val_dataset)\n        print(f\"Validation accuracy: {round(accuracy * 100, 2)}%\")\n    \n\n","metadata":{"execution":{"iopub.status.busy":"2022-03-17T11:45:44.483131Z","iopub.execute_input":"2022-03-17T11:45:44.483398Z","iopub.status.idle":"2022-03-17T11:45:44.496600Z","shell.execute_reply.started":"2022-03-17T11:45:44.483365Z","shell.execute_reply":"2022-03-17T11:45:44.495884Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"**We would like to see the images related to individual ids, so taken the samples base on  GroupKFold here.**\n\n*[Please Refer here for more inforamtion on GroupKfold](https://www.kaggle.com/reighns/groupkfold-and-stratified-groupkfold-efficientnet)*","metadata":{}},{"cell_type":"code","source":"from sklearn.model_selection import StratifiedKFold, GroupKFold\n\n# OOF RMSE Placeholder\n\nkfold = GroupKFold(n_splits = NUM_FOLDS)\nfor fold, (train_index, val_index) in enumerate(kfold.split(train_df.index, train_df['target_value'], groups=individual_id_group)):\n    mobilevit_xxs = create_mobilevit(num_classes=N_CLASSES)    \n    \n    run_experiment(mobilevit_xxs, fold)    ","metadata":{"execution":{"iopub.status.busy":"2022-03-17T11:45:47.457178Z","iopub.execute_input":"2022-03-17T11:45:47.457888Z","iopub.status.idle":"2022-03-17T12:52:55.901368Z","shell.execute_reply.started":"2022-03-17T11:45:47.457827Z","shell.execute_reply":"2022-03-17T12:52:55.900608Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"'''\nplt.figure(figsize=(8,8), tight_layout=True)\nplt.subplot(211)\nplt.plot(history.history['loss'])\nplt.plot(history.history['val_loss'])\nplt.title(' Loss')\nplt.ylabel('Error')\nplt.xlabel('Epoch')\nplt.legend(['train', 'val'], loc='upper right')\n\nplt.subplot(212)\nplt.plot(history.history['sparse_categorical_accuracy'])\nplt.plot(history.history['val_sparse_categorical_accuracy'])\nplt.title('SparseCategoricalAccuracy Metric')\nplt.ylabel('Error')\nplt.xlabel('Epoch')\nplt.legend(['train', 'val'], loc='upper left')\nplt.show()\n'''","metadata":{"_kg_hide-input":true,"_kg_hide-output":true,"execution":{"iopub.status.busy":"2022-03-17T10:38:29.508871Z","iopub.execute_input":"2022-03-17T10:38:29.509478Z","iopub.status.idle":"2022-03-17T10:38:29.516029Z","shell.execute_reply.started":"2022-03-17T10:38:29.509436Z","shell.execute_reply":"2022-03-17T10:38:29.515247Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Serialize the model as a SavedModel.\nmobilevit_xxs.save(\"mobilevit_xxs\")","metadata":{"execution":{"iopub.status.busy":"2022-03-17T09:41:20.014889Z","iopub.execute_input":"2022-03-17T09:41:20.015324Z","iopub.status.idle":"2022-03-17T09:41:47.939925Z","shell.execute_reply.started":"2022-03-17T09:41:20.015289Z","shell.execute_reply":"2022-03-17T09:41:47.939073Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# Inference","metadata":{}},{"cell_type":"code","source":"test_dataset = create_dataset(test_df,\n                                  batch_size  = BATCH_SIZE, \n                                  is_labelled = False, \n                                  augment = False,\n                                  repeat  = False, \n                                  shuffle = False,\n                            size = (image_size, image_size))\ntest_images = next(iter(test_dataset))\n","metadata":{"execution":{"iopub.status.busy":"2022-03-17T12:56:22.669310Z","iopub.execute_input":"2022-03-17T12:56:22.669803Z","iopub.status.idle":"2022-03-17T12:56:23.900957Z","shell.execute_reply.started":"2022-03-17T12:56:22.669764Z","shell.execute_reply":"2022-03-17T12:56:23.899736Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"plt.figure(figsize=(16, 10))\nfor i, image in enumerate(test_images[:20]):\n    ax = plt.subplot(5, 4,  i + 1)\n    #plt.title(\" Image\")\n    plt.imshow(image.numpy().squeeze())\n    plt.axis(\"off\")\n\nplt.tight_layout()\n","metadata":{"execution":{"iopub.status.busy":"2022-03-17T12:56:27.523202Z","iopub.execute_input":"2022-03-17T12:56:27.523496Z","iopub.status.idle":"2022-03-17T12:56:28.836327Z","shell.execute_reply.started":"2022-03-17T12:56:27.523462Z","shell.execute_reply":"2022-03-17T12:56:28.835707Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#batch = next(iter(test_dataset))\n","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"dict_ids =dict(zip( train_df['target_value'], train_df['individual_id']))","metadata":{"_kg_hide-input":true,"_kg_hide-output":true,"execution":{"iopub.status.busy":"2022-03-17T12:56:32.823228Z","iopub.execute_input":"2022-03-17T12:56:32.824786Z","iopub.status.idle":"2022-03-17T12:56:32.846158Z","shell.execute_reply.started":"2022-03-17T12:56:32.824731Z","shell.execute_reply":"2022-03-17T12:56:32.845283Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"test_preds = mobilevit_xxs.predict(test_dataset)","metadata":{"execution":{"iopub.status.busy":"2022-03-17T12:56:35.068830Z","iopub.execute_input":"2022-03-17T12:56:35.069093Z","iopub.status.idle":"2022-03-17T13:05:58.913741Z","shell.execute_reply.started":"2022-03-17T12:56:35.069061Z","shell.execute_reply":"2022-03-17T13:05:58.912757Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#dict_ids.get(1104)","metadata":{"execution":{"iopub.status.busy":"2022-03-17T10:17:13.115494Z","iopub.execute_input":"2022-03-17T10:17:13.116313Z","iopub.status.idle":"2022-03-17T10:17:13.122217Z","shell.execute_reply.started":"2022-03-17T10:17:13.116262Z","shell.execute_reply":"2022-03-17T10:17:13.121389Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"'''\nfor i in range(2):\n    arr = test_preds[i].argsort()[-top_n:][::-1]\n    print (arr)\n    r  = [''.join(dict_ids.get(arr[j])) for j in range(top_n)]\n    print (r)\n'''","metadata":{"execution":{"iopub.status.busy":"2022-03-17T13:17:31.072520Z","iopub.execute_input":"2022-03-17T13:17:31.072783Z","iopub.status.idle":"2022-03-17T13:17:31.083058Z","shell.execute_reply.started":"2022-03-17T13:17:31.072749Z","shell.execute_reply":"2022-03-17T13:17:31.082235Z"},"_kg_hide-input":true,"_kg_hide-output":true,"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"top_n = 5\nfor i in range(len(test_preds)):\n    arr = test_preds[i].argsort()[-top_n:][::-1]\n    #print (arr)\n    test_df['predictions'][i]  = [''.join(dict_ids.get(arr[j])) for j in range(top_n)]\n    #print (r)\n    #test_df['predictions'][i] = r\n","metadata":{"execution":{"iopub.status.busy":"2022-03-17T13:18:22.003619Z","iopub.execute_input":"2022-03-17T13:18:22.004421Z","iopub.status.idle":"2022-03-17T13:19:09.846972Z","shell.execute_reply.started":"2022-03-17T13:18:22.004379Z","shell.execute_reply":"2022-03-17T13:19:09.846147Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"sub_df = test_df[['image', 'predictions']]","metadata":{"execution":{"iopub.status.busy":"2022-03-17T13:20:00.272785Z","iopub.execute_input":"2022-03-17T13:20:00.273448Z","iopub.status.idle":"2022-03-17T13:20:00.282131Z","shell.execute_reply.started":"2022-03-17T13:20:00.273406Z","shell.execute_reply":"2022-03-17T13:20:00.281216Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"sub_df.to_csv('submission.csv', index = False)","metadata":{"execution":{"iopub.status.busy":"2022-03-17T13:20:08.360198Z","iopub.execute_input":"2022-03-17T13:20:08.360487Z","iopub.status.idle":"2022-03-17T13:20:08.513854Z","shell.execute_reply.started":"2022-03-17T13:20:08.360457Z","shell.execute_reply":"2022-03-17T13:20:08.513060Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"!tail submission.csv","metadata":{"execution":{"iopub.status.busy":"2022-03-17T13:20:13.040267Z","iopub.execute_input":"2022-03-17T13:20:13.040614Z","iopub.status.idle":"2022-03-17T13:20:13.838153Z","shell.execute_reply.started":"2022-03-17T13:20:13.040576Z","shell.execute_reply":"2022-03-17T13:20:13.837321Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"Provide your feedback to help me improve further. Thank you.","metadata":{}}]}