{"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":"# 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\n#import numpy as np # linear algebra\n#import 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\n#import os\n#for 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","execution":{"iopub.status.busy":"2023-05-06T10:34:34.518597Z","iopub.execute_input":"2023-05-06T10:34:34.519315Z","iopub.status.idle":"2023-05-06T10:34:34.544954Z","shell.execute_reply.started":"2023-05-06T10:34:34.519276Z","shell.execute_reply":"2023-05-06T10:34:34.543924Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import os, json, random, math, scipy, wandb\nimport numpy as np\nimport pandas as pd\nimport tensorflow as tf\nimport tensorflow_addons as tfa\nimport matplotlib.pyplot as plt\nfrom tqdm.notebook import tqdm\nfrom sklearn.model_selection import StratifiedGroupKFold \nfrom types import SimpleNamespace\nfrom pathlib import Path","metadata":{"execution":{"iopub.status.busy":"2023-05-06T10:34:34.546656Z","iopub.execute_input":"2023-05-06T10:34:34.546977Z","iopub.status.idle":"2023-05-06T10:34:43.557798Z","shell.execute_reply.started":"2023-05-06T10:34:34.546947Z","shell.execute_reply":"2023-05-06T10:34:43.556703Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"try:\n    from kaggle_secrets import UserSecretsClient\n    user_secrets = UserSecretsClient()\n    secret_value_0 = user_secrets.get_secret(\"82c0e9e68dd26ea25025b0f01592b0ed1c13c87b\")\n    wandb.login(key=secret_value_0)\nexcept:\n    wandb.login()","metadata":{"execution":{"iopub.status.busy":"2023-05-06T10:34:43.559287Z","iopub.execute_input":"2023-05-06T10:34:43.560199Z","iopub.status.idle":"2023-05-06T10:34:51.532349Z","shell.execute_reply.started":"2023-05-06T10:34:43.560159Z","shell.execute_reply":"2023-05-06T10:34:51.531219Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Source: https://www.kaggle.com/competitions/asl-signs/overview/evaluation\nROWS_PER_FRAME = 543\n\ndef load_relevant_data_subset(pq_path):\n    data_columns = ['x', 'y', 'z']\n    data = pd.read_parquet(pq_path, columns=data_columns)\n    n_frames = int(len(data) / ROWS_PER_FRAME)\n    data = data.values.reshape(n_frames, ROWS_PER_FRAME, len(data_columns))\n    return data.astype(np.float32)","metadata":{"execution":{"iopub.status.busy":"2023-05-06T10:34:51.535099Z","iopub.execute_input":"2023-05-06T10:34:51.535479Z","iopub.status.idle":"2023-05-06T10:34:51.542206Z","shell.execute_reply.started":"2023-05-06T10:34:51.535436Z","shell.execute_reply":"2023-05-06T10:34:51.540823Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Config and Setup","metadata":{}},{"cell_type":"code","source":"cfg = SimpleNamespace()","metadata":{"execution":{"iopub.status.busy":"2023-05-06T10:34:51.543973Z","iopub.execute_input":"2023-05-06T10:34:51.544769Z","iopub.status.idle":"2023-05-06T10:34:51.555241Z","shell.execute_reply.started":"2023-05-06T10:34:51.544728Z","shell.execute_reply":"2023-05-06T10:34:51.554226Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"iskaggle = os.environ.get('KAGGLE_KERNEL_RUN_TYPE', '')\niskaggle","metadata":{"execution":{"iopub.status.busy":"2023-05-06T10:34:51.556945Z","iopub.execute_input":"2023-05-06T10:34:51.557352Z","iopub.status.idle":"2023-05-06T10:34:51.570342Z","shell.execute_reply.started":"2023-05-06T10:34:51.557316Z","shell.execute_reply":"2023-05-06T10:34:51.569362Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"DATA_DIR         = Path('../data/') if not iskaggle else Path('/kaggle/input/asl-signs/')\nTRAIN_CSV_PATH   = DATA_DIR/'train.csv'\nLANDMARK_DIR     = DATA_DIR/'train_landmark_files'\nLABEL_MAP_PATH   = DATA_DIR/'sign_to_prediction_index_map.json'\nDATA_DIR","metadata":{"execution":{"iopub.status.busy":"2023-05-06T10:34:51.572117Z","iopub.execute_input":"2023-05-06T10:34:51.572411Z","iopub.status.idle":"2023-05-06T10:34:51.583927Z","shell.execute_reply.started":"2023-05-06T10:34:51.572385Z","shell.execute_reply":"2023-05-06T10:34:51.582976Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"cfg.PREPROCESS_DATA = False\ncfg.TRAIN_MODEL = True\ncfg.N_ROWS = 543\ncfg.N_DIMS = 3\ncfg.DIM_NAMES = ['x', 'y', 'z']\ncfg.SEED = 42\ncfg.NUM_CLASSES = 250\ncfg.IS_INTERACTIVE = True\ncfg.VERBOSE = 2\ncfg.INPUT_SIZE = 32\ncfg.BATCH_ALL_SIGNS_N = 4\ncfg.BATCH_SIZE = 256\ncfg.N_EPOCHS = 50\ncfg.LR_MAX = 1e-3\ncfg.N_WARMUP_EPOCHS = 0\ncfg.WD_RATIO = 0.05\ncfg.MASK_VAL = 4237","metadata":{"execution":{"iopub.status.busy":"2023-05-06T10:34:51.585727Z","iopub.execute_input":"2023-05-06T10:34:51.586225Z","iopub.status.idle":"2023-05-06T10:34:51.595635Z","shell.execute_reply.started":"2023-05-06T10:34:51.586190Z","shell.execute_reply":"2023-05-06T10:34:51.594627Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train = pd.read_csv(TRAIN_CSV_PATH)\nN_SAMPLES = len(train)\nN_SAMPLES","metadata":{"execution":{"iopub.status.busy":"2023-05-06T10:34:51.600671Z","iopub.execute_input":"2023-05-06T10:34:51.600949Z","iopub.status.idle":"2023-05-06T10:34:51.838805Z","shell.execute_reply.started":"2023-05-06T10:34:51.600924Z","shell.execute_reply":"2023-05-06T10:34:51.837833Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"#### Below, instead of updating the path on the train file, we create a symlink instead.","metadata":{}},{"cell_type":"code","source":"# Get complete file path to file\ndef get_file_path(path):\n    return f'/kaggle/input/asl-signs/{path}'\n\n!ln -s {LANDMARK_DIR} ./train_landmark_files\ntrain['file_path'] = train['path'].values\ntrain.head(2)","metadata":{"execution":{"iopub.status.busy":"2023-05-06T10:34:51.844503Z","iopub.execute_input":"2023-05-06T10:34:51.850137Z","iopub.status.idle":"2023-05-06T10:34:52.877066Z","shell.execute_reply.started":"2023-05-06T10:34:51.850073Z","shell.execute_reply":"2023-05-06T10:34:52.875755Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"with open(LABEL_MAP_PATH, 'r') as f:\n    data = json.load(f)\n\n# Create a dictionary to map the indexes to their corresponding values\nindex_to_value = {str(index): value for index, value in data.items()}\n\n# add a new column with the corresponding keys using map\ntrain['sign_ord'] = train['sign'].map(index_to_value)\ntrain.head(2)","metadata":{"execution":{"iopub.status.busy":"2023-05-06T10:34:52.879196Z","iopub.execute_input":"2023-05-06T10:34:52.879593Z","iopub.status.idle":"2023-05-06T10:34:52.906816Z","shell.execute_reply.started":"2023-05-06T10:34:52.879546Z","shell.execute_reply":"2023-05-06T10:34:52.905716Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Kfold","metadata":{}},{"cell_type":"markdown","source":"Below, we create K-folds using StratifiedGroupKFold, the idea is to have different participants in\ntrain and validation.","metadata":{}},{"cell_type":"code","source":"from sklearn.model_selection import StratifiedGroupKFold","metadata":{"execution":{"iopub.status.busy":"2023-05-06T10:34:52.908566Z","iopub.execute_input":"2023-05-06T10:34:52.908963Z","iopub.status.idle":"2023-05-06T10:34:52.913922Z","shell.execute_reply.started":"2023-05-06T10:34:52.908926Z","shell.execute_reply":"2023-05-06T10:34:52.912676Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"N_PARTICIPANTS = train.participant_id.nunique()\nsgkf = StratifiedGroupKFold(n_splits=7, shuffle=True, random_state=43)\ntrain['fold'] = -1\nfor i, (train_idx, val_idx) in enumerate(sgkf.split(train.index, train.sign, train.participant_id)):\n    train.loc[val_idx, 'fold'] = i\ntrain.head()","metadata":{"execution":{"iopub.status.busy":"2023-05-06T10:34:52.915741Z","iopub.execute_input":"2023-05-06T10:34:52.916557Z","iopub.status.idle":"2023-05-06T10:34:53.222347Z","shell.execute_reply.started":"2023-05-06T10:34:52.916520Z","shell.execute_reply":"2023-05-06T10:34:53.221288Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# create indexes using fold `0` for now\ntrain_idxs = train.query(\"fold!=0\").index.values\nval_idxs = train.query(\"fold==0\").index.values","metadata":{"execution":{"iopub.status.busy":"2023-05-06T10:34:53.224041Z","iopub.execute_input":"2023-05-06T10:34:53.224449Z","iopub.status.idle":"2023-05-06T10:34:53.255805Z","shell.execute_reply.started":"2023-05-06T10:34:53.224407Z","shell.execute_reply":"2023-05-06T10:34:53.254879Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"data = pd.read_parquet(train.loc[0, \"file_path\"])\ndata[data['frame'] == 20]","metadata":{"execution":{"iopub.status.busy":"2023-05-06T10:34:53.257216Z","iopub.execute_input":"2023-05-06T10:34:53.257796Z","iopub.status.idle":"2023-05-06T10:34:53.368087Z","shell.execute_reply.started":"2023-05-06T10:34:53.257742Z","shell.execute_reply":"2023-05-06T10:34:53.367092Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Process Data Tensorflow\n","metadata":{}},{"cell_type":"code","source":"# landmark indices in original data\nLIPS_IDXS0 = np.array([\n        61, 185, 40, 39, 37, 0, 267, 269, 270, 409,\n        291, 146, 91, 181, 84, 17, 314, 405, 321, 375,\n        78, 191, 80, 81, 82, 13, 312, 311, 310, 415,\n        95, 88, 178, 87, 14, 317, 402, 318, 324, 308,\n    ])\nLEFT_HAND_IDXS0  = np.arange(468,489)\nRIGHT_HAND_IDXS0 = np.arange(522,543)\nPOSE_IDXS0       = np.arange(502, 512)\nLANDMARK_IDXS0   = np.concatenate((LIPS_IDXS0, LEFT_HAND_IDXS0, RIGHT_HAND_IDXS0, POSE_IDXS0))\nHAND_IDXS0       = np.concatenate((LEFT_HAND_IDXS0, RIGHT_HAND_IDXS0), axis=0)\nN_COLS           = LANDMARK_IDXS0.size\nN_COLS, LANDMARK_IDXS0","metadata":{"execution":{"iopub.status.busy":"2023-05-06T10:34:53.369533Z","iopub.execute_input":"2023-05-06T10:34:53.370165Z","iopub.status.idle":"2023-05-06T10:34:53.381929Z","shell.execute_reply.started":"2023-05-06T10:34:53.370125Z","shell.execute_reply":"2023-05-06T10:34:53.380712Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Landmark indices in processed data\nLIPS_IDXS       = np.argwhere(np.isin(LANDMARK_IDXS0, LIPS_IDXS0)).squeeze()\nLEFT_HAND_IDXS  = np.argwhere(np.isin(LANDMARK_IDXS0, LEFT_HAND_IDXS0)).squeeze()\nRIGHT_HAND_IDXS = np.argwhere(np.isin(LANDMARK_IDXS0, RIGHT_HAND_IDXS0)).squeeze()\nHAND_IDXS       = np.argwhere(np.isin(LANDMARK_IDXS0, HAND_IDXS0)).squeeze()\nPOSE_IDXS       = np.argwhere(np.isin(LANDMARK_IDXS0, POSE_IDXS0)).squeeze()\nHAND_IDXS","metadata":{"execution":{"iopub.status.busy":"2023-05-06T10:34:53.383177Z","iopub.execute_input":"2023-05-06T10:34:53.384241Z","iopub.status.idle":"2023-05-06T10:34:53.396951Z","shell.execute_reply.started":"2023-05-06T10:34:53.384203Z","shell.execute_reply":"2023-05-06T10:34:53.396071Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"class PreprocessLayer(tf.keras.layers.Layer):\n    def __init__(self):\n        super(PreprocessLayer, self).__init__()\n        \n    def pad_edge(self, t, repeats, side):\n        if side == 'LEFT':\n            return tf.concat((tf.repeat(t[:1], repeats=repeats, axis=0), t), axis=0)\n        elif side == 'RIGHT':\n            return tf.concat((t, tf.repeat(t[-1:], repeats=repeats, axis=0)), axis=0)\n    \n    @tf.function(\n        input_signature=(tf.TensorSpec(shape=[None,cfg.N_ROWS,cfg.N_DIMS], dtype=tf.float32),),\n    )\n    def call(self, data0):\n        # Number of Frames in Video\n        N_FRAMES0 = tf.shape(data0)[0]\n        \n        # Keep only non-empty frames in data\n        frames_hands_nansum = tf.experimental.numpy.nanmean(tf.gather(data0, HAND_IDXS0, axis=1), axis=[1,2])\n        non_empty_frames_idxs = tf.where(frames_hands_nansum > 0)\n        non_empty_frames_idxs = tf.squeeze(non_empty_frames_idxs, axis=1)\n        data = tf.gather(data0, non_empty_frames_idxs, axis=0)\n        \n        non_empty_frames_idxs = tf.cast(non_empty_frames_idxs, tf.float32) \n        \n        # Number of non-empty frames\n        N_FRAMES = tf.shape(data)[0]\n        data = tf.gather(data, LANDMARK_IDXS0, axis=1)\n        \n        if N_FRAMES < cfg.INPUT_SIZE:\n            # Video fits in cfg.INPUT_SIZE\n            non_empty_frames_idxs = tf.pad(non_empty_frames_idxs, [[0, cfg.INPUT_SIZE-N_FRAMES]], constant_values=-1)\n            data = tf.pad(data, [[0, cfg.INPUT_SIZE-N_FRAMES], [0,0], [0,0]], constant_values=0)\n            data = tf.where(tf.math.is_nan(data), 0.0, data)\n            return data, non_empty_frames_idxs\n        else:\n            # Video needs to be downsampled to cfg.INPUT_SIZE\n            if N_FRAMES < cfg.INPUT_SIZE**2:\n                repeats = tf.math.floordiv(cfg.INPUT_SIZE * cfg.INPUT_SIZE, N_FRAMES0)\n                data = tf.repeat(data, repeats=repeats, axis=0)\n                non_empty_frames_idxs = tf.repeat(non_empty_frames_idxs, repeats=repeats, axis=0)\n\n            # Pad To Multiple Of Input Size\n            pool_size = tf.math.floordiv(len(data), cfg.INPUT_SIZE)\n            if tf.math.mod(len(data), cfg.INPUT_SIZE) > 0:\n                pool_size += 1\n            if pool_size == 1:\n                pad_size = (pool_size * cfg.INPUT_SIZE) - len(data)\n            else:\n                pad_size = (pool_size * cfg.INPUT_SIZE) % len(data)\n\n            # Pad Start/End with Start/End value\n            pad_left = tf.math.floordiv(pad_size, 2) + tf.math.floordiv(cfg.INPUT_SIZE, 2)\n            pad_right = tf.math.floordiv(pad_size, 2) + tf.math.floordiv(cfg.INPUT_SIZE, 2)\n            if tf.math.mod(pad_size, 2) > 0:\n                pad_right += 1\n\n            # Pad By Concatenating Left/Right Edge Values\n            data = self.pad_edge(data, pad_left, 'LEFT')\n            data = self.pad_edge(data, pad_right, 'RIGHT')\n\n            # Pad Non Empty Frame Indices\n            non_empty_frames_idxs = self.pad_edge(non_empty_frames_idxs, pad_left, 'LEFT')\n            non_empty_frames_idxs = self.pad_edge(non_empty_frames_idxs, pad_right, 'RIGHT')\n\n            # Reshape to Mean Pool\n            data = tf.reshape(data, [cfg.INPUT_SIZE, -1, N_COLS, cfg.N_DIMS])\n            non_empty_frames_idxs = tf.reshape(non_empty_frames_idxs, [cfg.INPUT_SIZE, -1])\n\n            # Mean Pool\n            data = tf.experimental.numpy.nanmean(data, axis=1)\n            non_empty_frames_idxs = tf.experimental.numpy.nanmean(non_empty_frames_idxs, axis=1)\n\n            # Fill NaN Values With 0\n            data = tf.where(tf.math.is_nan(data), 0.0, data)\n            \n            return data, non_empty_frames_idxs\n    \npreprocess_layer = PreprocessLayer()","metadata":{"execution":{"iopub.status.busy":"2023-05-06T10:34:53.398502Z","iopub.execute_input":"2023-05-06T10:34:53.399228Z","iopub.status.idle":"2023-05-06T10:34:53.428813Z","shell.execute_reply.started":"2023-05-06T10:34:53.399191Z","shell.execute_reply":"2023-05-06T10:34:53.427831Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"sample = load_relevant_data_subset(train.path[2])\nsample.shape","metadata":{"execution":{"iopub.status.busy":"2023-05-06T10:34:53.430601Z","iopub.execute_input":"2023-05-06T10:34:53.431408Z","iopub.status.idle":"2023-05-06T10:34:53.466450Z","shell.execute_reply.started":"2023-05-06T10:34:53.431371Z","shell.execute_reply":"2023-05-06T10:34:53.465823Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"data, non_empty_frames_idxs = preprocess_layer(sample)\ndata","metadata":{"execution":{"iopub.status.busy":"2023-05-06T10:34:53.467541Z","iopub.execute_input":"2023-05-06T10:34:53.468493Z","iopub.status.idle":"2023-05-06T10:34:57.014792Z","shell.execute_reply.started":"2023-05-06T10:34:53.468456Z","shell.execute_reply":"2023-05-06T10:34:57.013773Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"LIPS_START = 0\nLEFT_HAND_START = LIPS_IDXS.size\nRIGHT_HAND_START = LEFT_HAND_START + LEFT_HAND_IDXS.size\nPOSE_START = RIGHT_HAND_START + RIGHT_HAND_IDXS.size","metadata":{"execution":{"iopub.status.busy":"2023-05-06T10:34:57.016480Z","iopub.execute_input":"2023-05-06T10:34:57.016865Z","iopub.status.idle":"2023-05-06T10:34:57.024162Z","shell.execute_reply.started":"2023-05-06T10:34:57.016827Z","shell.execute_reply":"2023-05-06T10:34:57.023189Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def get_data(file_path):\n    data = load_relevant_data_subset(file_path)\n    data = preprocess_layer(data)\n    return data","metadata":{"execution":{"iopub.status.busy":"2023-05-06T10:34:57.026029Z","iopub.execute_input":"2023-05-06T10:34:57.026442Z","iopub.status.idle":"2023-05-06T10:34:57.033958Z","shell.execute_reply.started":"2023-05-06T10:34:57.026404Z","shell.execute_reply":"2023-05-06T10:34:57.032844Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def get_x_y():\n    # Create arrays to save data\n    X = np.zeros([N_SAMPLES, cfg.INPUT_SIZE, N_COLS, cfg.N_DIMS], dtype=np.float32)\n    y = np.zeros([N_SAMPLES], dtype=np.int32)\n    NON_EMPTY_FRAME_IDXS = np.full([N_SAMPLES, cfg.INPUT_SIZE], -1, dtype=np.float32)\n\n    for row_idx, (file_path, sign_ord) in enumerate(tqdm(train[['file_path', 'sign_ord']].values)):\n        if row_idx % 5000 == 0:\n            print(f'Generated {row_idx}/{N_SAMPLES}')\n\n        data, non_empty_frame_idxs = get_data(file_path)\n        X[row_idx] = data\n        y[row_idx] = sign_ord\n        NON_EMPTY_FRAME_IDXS[row_idx] = non_empty_frame_idxs\n        if np.isnan(data).sum() > 0: return data\n\n    # Save X/y\n    np.save('X.npy', X)\n    np.save('y.npy', y)\n    np.save('NON_EMPTY_FRAME_IDXS.npy', NON_EMPTY_FRAME_IDXS)\n    \n    return X, y, NON_EMPTY_FRAME_IDXS","metadata":{"execution":{"iopub.status.busy":"2023-05-06T10:34:57.035472Z","iopub.execute_input":"2023-05-06T10:34:57.036286Z","iopub.status.idle":"2023-05-06T10:34:57.049111Z","shell.execute_reply.started":"2023-05-06T10:34:57.036250Z","shell.execute_reply":"2023-05-06T10:34:57.048125Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"if not cfg.PREPROCESS_DATA:\n    X, y, NON_EMPTY_FRAME_IDXS = get_x_y()\nelse:\n    X = np.load('/kaggle/working/X.npy')\n    y = np.load('/kaggle/working/y.npy')\n    NON_EMPTY_FRAME_IDXS = np.load('/kaggle/working/NON_EMPTY_FRAME_IDXS.npy')","metadata":{"execution":{"iopub.status.busy":"2023-05-06T10:34:57.056012Z","iopub.execute_input":"2023-05-06T10:34:57.056290Z","iopub.status.idle":"2023-05-06T11:04:47.584161Z","shell.execute_reply.started":"2023-05-06T10:34:57.056255Z","shell.execute_reply":"2023-05-06T11:04:47.583013Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"X.shape, y.shape, NON_EMPTY_FRAME_IDXS.shape","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:04:47.586676Z","iopub.execute_input":"2023-05-06T11:04:47.588556Z","iopub.status.idle":"2023-05-06T11:04:47.597548Z","shell.execute_reply.started":"2023-05-06T11:04:47.588470Z","shell.execute_reply":"2023-05-06T11:04:47.596425Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Feature Statistics - Lips\n","metadata":{}},{"cell_type":"code","source":"# LIPS\nLIPS_MEAN_X  = np.zeros([LIPS_IDXS.size], dtype=np.float32)\nLIPS_MEAN_Y  = np.zeros([LIPS_IDXS.size], dtype=np.float32)\nLIPS_STD_X   = np.zeros([LIPS_IDXS.size], dtype=np.float32)\nLIPS_STD_Y   = np.zeros([LIPS_IDXS.size], dtype=np.float32)\n\nfor col, ll in enumerate(tqdm( np.transpose(X[:,:,LIPS_IDXS], [2,3,0,1]).reshape([LIPS_IDXS.size, cfg.N_DIMS, -1]) )):\n    for dim, l in enumerate(ll):\n        v = l[np.nonzero(l)]\n        if dim == 0: # X\n            LIPS_MEAN_X[col] = v.mean()\n            LIPS_STD_X[col] = v.std()\n        if dim == 1: # Y\n            LIPS_MEAN_Y[col] = v.mean()\n            LIPS_STD_Y[col] = v.std()\n        \nLIPS_MEAN = np.array([LIPS_MEAN_X, LIPS_MEAN_Y]).T\nLIPS_STD = np.array([LIPS_STD_X, LIPS_STD_Y]).T","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:04:47.599640Z","iopub.execute_input":"2023-05-06T11:04:47.600018Z","iopub.status.idle":"2023-05-06T11:04:58.331927Z","shell.execute_reply.started":"2023-05-06T11:04:47.599979Z","shell.execute_reply":"2023-05-06T11:04:58.330848Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Feature Statistics - Hands","metadata":{}},{"cell_type":"code","source":"# LEFT HAND\nLEFT_HANDS_MEAN_X = np.zeros([LEFT_HAND_IDXS.size], dtype=np.float32)\nLEFT_HANDS_MEAN_Y = np.zeros([LEFT_HAND_IDXS.size], dtype=np.float32)\nLEFT_HANDS_STD_X = np.zeros([LEFT_HAND_IDXS.size], dtype=np.float32)\nLEFT_HANDS_STD_Y = np.zeros([LEFT_HAND_IDXS.size], dtype=np.float32)\n# RIGHT HAND\nRIGHT_HANDS_MEAN_X = np.zeros([RIGHT_HAND_IDXS.size], dtype=np.float32)\nRIGHT_HANDS_MEAN_Y = np.zeros([RIGHT_HAND_IDXS.size], dtype=np.float32)\nRIGHT_HANDS_STD_X = np.zeros([RIGHT_HAND_IDXS.size], dtype=np.float32)\nRIGHT_HANDS_STD_Y = np.zeros([RIGHT_HAND_IDXS.size], dtype=np.float32)\n\nfor col, ll in enumerate(tqdm( np.transpose(X[:,:,HAND_IDXS], [2,3,0,1]).reshape([HAND_IDXS.size, cfg.N_DIMS, -1]) )):\n    for dim, l in enumerate(ll):\n        v = l[np.nonzero(l)]\n        if dim == 0: # X\n            if col < RIGHT_HAND_IDXS.size: # LEFT HAND\n                LEFT_HANDS_MEAN_X[col] = v.mean()\n                LEFT_HANDS_STD_X[col] = v.std()\n            else:\n                RIGHT_HANDS_MEAN_X[col - LEFT_HAND_IDXS.size] = v.mean()\n                RIGHT_HANDS_STD_X[col - LEFT_HAND_IDXS.size] = v.std()\n        if dim == 1: # Y\n            if col < RIGHT_HAND_IDXS.size: # LEFT HAND\n                LEFT_HANDS_MEAN_Y[col] = v.mean()\n                LEFT_HANDS_STD_Y[col] = v.std()\n            else: # RIGHT HAND\n                RIGHT_HANDS_MEAN_Y[col - LEFT_HAND_IDXS.size] = v.mean()\n                RIGHT_HANDS_STD_Y[col - LEFT_HAND_IDXS.size] = v.std()\n        \nLEFT_HANDS_MEAN = np.array([LEFT_HANDS_MEAN_X, LEFT_HANDS_MEAN_Y]).T\nLEFT_HANDS_STD = np.array([LEFT_HANDS_STD_X, LEFT_HANDS_STD_Y]).T\nRIGHT_HANDS_MEAN = np.array([RIGHT_HANDS_MEAN_X, RIGHT_HANDS_MEAN_Y]).T\nRIGHT_HANDS_STD = np.array([RIGHT_HANDS_STD_X, RIGHT_HANDS_STD_Y]).T","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:04:58.333328Z","iopub.execute_input":"2023-05-06T11:04:58.334307Z","iopub.status.idle":"2023-05-06T11:05:08.178963Z","shell.execute_reply.started":"2023-05-06T11:04:58.334267Z","shell.execute_reply":"2023-05-06T11:05:08.177845Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Feature Statistics - Pose","metadata":{}},{"cell_type":"code","source":"# POSE\nPOSE_MEAN_X = np.zeros([POSE_IDXS.size], dtype=np.float32)\nPOSE_MEAN_Y = np.zeros([POSE_IDXS.size], dtype=np.float32)\nPOSE_STD_X = np.zeros([POSE_IDXS.size], dtype=np.float32)\nPOSE_STD_Y = np.zeros([POSE_IDXS.size], dtype=np.float32)\n\nfor col, ll in enumerate(tqdm( np.transpose(X[:,:,POSE_IDXS], [2,3,0,1]).reshape([POSE_IDXS.size, cfg.N_DIMS, -1]) )):\n    for dim, l in enumerate(ll):\n        v = l[np.nonzero(l)]\n        if dim == 0: # X\n            POSE_MEAN_X[col] = v.mean()\n            POSE_STD_X[col] = v.std()\n        if dim == 1: # Y\n            POSE_MEAN_Y[col] = v.mean()\n            POSE_STD_Y[col] = v.std()\n        \nPOSE_MEAN = np.array([POSE_MEAN_X, POSE_MEAN_Y]).T\nPOSE_STD = np.array([POSE_STD_X, POSE_STD_Y]).T","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:05:08.180576Z","iopub.execute_input":"2023-05-06T11:05:08.181239Z","iopub.status.idle":"2023-05-06T11:05:10.171142Z","shell.execute_reply.started":"2023-05-06T11:05:08.181197Z","shell.execute_reply":"2023-05-06T11:05:10.169843Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Custom sampler to get a batch containing N times all signs\ndef get_train_batch_all_signs(X, y, NON_EMPTY_FRAME_IDXS, n=cfg.BATCH_ALL_SIGNS_N):\n    # Arrays to store batch in\n    X_batch = np.zeros([cfg.NUM_CLASSES*n, cfg.INPUT_SIZE, N_COLS, cfg.N_DIMS], dtype=np.float32)\n    y_batch = np.arange(0, cfg.NUM_CLASSES, step=1/n, dtype=np.float32).astype(np.int64)\n    non_empty_frame_idxs_batch = np.zeros([cfg.NUM_CLASSES*n, cfg.INPUT_SIZE], dtype=np.float32)\n    \n    # Dictionary mapping ordinally encoded sign to corresponding sample indices\n    CLASS2IDXS = {}\n    for i in range(cfg.NUM_CLASSES):\n        CLASS2IDXS[i] = np.argwhere(y == i).squeeze().astype(np.int32)\n            \n    while True:\n        # Fill batch arrays\n        for i in range(cfg.NUM_CLASSES):\n            idxs = np.random.choice(CLASS2IDXS[i], n)\n            X_batch[i*n:(i+1)*n] = X[idxs]\n            non_empty_frame_idxs_batch[i*n:(i+1)*n] = NON_EMPTY_FRAME_IDXS[idxs]\n        \n        yield { 'frames': X_batch, 'non_empty_frame_idxs': non_empty_frame_idxs_batch }, y_batch","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:05:10.172986Z","iopub.execute_input":"2023-05-06T11:05:10.173380Z","iopub.status.idle":"2023-05-06T11:05:10.185532Z","shell.execute_reply.started":"2023-05-06T11:05:10.173339Z","shell.execute_reply":"2023-05-06T11:05:10.184460Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"dummy_dataset = get_train_batch_all_signs(X, y, NON_EMPTY_FRAME_IDXS)\nX_batch, y_batch = next(dummy_dataset)\nX_batch.keys(), X_batch['frames'].shape","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:05:10.186969Z","iopub.execute_input":"2023-05-06T11:05:10.187661Z","iopub.status.idle":"2023-05-06T11:05:10.248808Z","shell.execute_reply.started":"2023-05-06T11:05:10.187621Z","shell.execute_reply":"2023-05-06T11:05:10.247692Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Model Config\n","metadata":{}},{"cell_type":"code","source":"# Epsilon value for layer normalisation\nLAYER_NORM_EPS = 1e-6\n\n# Dense layer units for landmarks\nLIPS_UNITS = 384\nHANDS_UNITS = 384\nPOSE_UNITS = 384\n# final embedding and transformer embedding size\nUNITS = 384\n\n# Transformer\nNUM_BLOCKS = 2\nMLP_RATIO = 2\n\n# Dropout\nEMBEDDING_DROPOUT = 0.00\nMLP_DROPOUT_RATIO = 0.30\nCLASSIFIER_DROPOUT_RATIO = 0.10\n\n# Initiailizers\nINIT_HE_UNIFORM = tf.keras.initializers.he_uniform\nINIT_GLOROT_UNIFORM = tf.keras.initializers.glorot_uniform\nINIT_ZEROS = tf.keras.initializers.constant(0.0)\n# Activations\nGELU = tf.keras.activations.gelu","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:05:10.250362Z","iopub.execute_input":"2023-05-06T11:05:10.250899Z","iopub.status.idle":"2023-05-06T11:05:10.257644Z","shell.execute_reply.started":"2023-05-06T11:05:10.250856Z","shell.execute_reply":"2023-05-06T11:05:10.256635Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Transformer\n\nNeed to implement transformer from scratch as TFLite does not support the native TF implementation of MultiHeadAttention.","metadata":{}},{"cell_type":"code","source":"def scaled_dot_product(q,k,v, softmax, attention_mask):\n    #calculates Q . K(transpose)\n    qkt = tf.matmul(q,k,transpose_b=True)\n    #caculates scaling factor\n    dk = tf.math.sqrt(tf.cast(q.shape[-1],dtype=tf.float32))\n    scaled_qkt = qkt/dk\n    softmax = softmax(scaled_qkt, mask=attention_mask)\n    \n    z = tf.matmul(softmax,v)\n    #shape: (m,Tx,depth), same shape as q,k,v\n    return z\n\nclass MultiHeadAttention(tf.keras.layers.Layer):\n    def __init__(self,d_model,num_of_heads):\n        super(MultiHeadAttention,self).__init__()\n        self.d_model = d_model\n        self.num_of_heads = num_of_heads\n        self.depth = d_model//num_of_heads\n        self.wq = [tf.keras.layers.Dense(self.depth) for i in range(num_of_heads)]\n        self.wk = [tf.keras.layers.Dense(self.depth) for i in range(num_of_heads)]\n        self.wv = [tf.keras.layers.Dense(self.depth) for i in range(num_of_heads)]\n        self.wo = tf.keras.layers.Dense(d_model)\n        self.softmax = tf.keras.layers.Softmax()\n        \n    def call(self,x, attention_mask):\n        \n        multi_attn = []\n        for i in range(self.num_of_heads):\n            Q = self.wq[i](x)\n            K = self.wk[i](x)\n            V = self.wv[i](x)\n            multi_attn.append(scaled_dot_product(Q,K,V, self.softmax, attention_mask))\n            \n        multi_head = tf.concat(multi_attn,axis=-1)\n        multi_head_attention = self.wo(multi_head)\n        return multi_head_attention","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:05:10.259272Z","iopub.execute_input":"2023-05-06T11:05:10.259958Z","iopub.status.idle":"2023-05-06T11:05:10.278477Z","shell.execute_reply.started":"2023-05-06T11:05:10.259918Z","shell.execute_reply":"2023-05-06T11:05:10.277324Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"class Transformer(tf.keras.Model):\n    def __init__(self, num_blocks):\n        super(Transformer, self).__init__(name='transformer')\n        self.num_blocks = num_blocks\n    \n    def build(self, input_shape):\n        self.ln_1s = []\n        self.mhas = []\n        self.ln_2s = []\n        self.mlps = []\n        # Make Transformer Blocks\n        for i in range(self.num_blocks):\n            # First Layer Normalisation\n            self.ln_1s.append(tf.keras.layers.LayerNormalization(epsilon=LAYER_NORM_EPS))\n            # Multi Head Attention\n            self.mhas.append(MultiHeadAttention(UNITS, 8))\n            # Second Layer Normalisation\n            self.ln_2s.append(tf.keras.layers.LayerNormalization(epsilon=LAYER_NORM_EPS))\n            # Multi Layer Perception\n            self.mlps.append(tf.keras.Sequential([\n                tf.keras.layers.Dense(UNITS * MLP_RATIO, activation=GELU, kernel_initializer=INIT_GLOROT_UNIFORM),\n                tf.keras.layers.Dropout(MLP_DROPOUT_RATIO),\n                tf.keras.layers.Dense(UNITS, kernel_initializer=INIT_HE_UNIFORM),\n            ]))\n        \n    def call(self, x, attention_mask):\n        # Iterate input over transformer blocks\n        for ln_1, mha, ln_2, mlp in zip(self.ln_1s, self.mhas, self.ln_2s, self.mlps):\n            x1 = ln_1(x)\n            attention_output = mha(x1, attention_mask)\n            x2 = x1 + attention_output\n            x3 = ln_2(x2)\n            x3 = mlp(x3)\n            x = x3 + x2\n    \n        return x","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:05:10.282760Z","iopub.execute_input":"2023-05-06T11:05:10.283559Z","iopub.status.idle":"2023-05-06T11:05:10.306505Z","shell.execute_reply.started":"2023-05-06T11:05:10.283514Z","shell.execute_reply":"2023-05-06T11:05:10.302279Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Landmark Embedding","metadata":{}},{"cell_type":"code","source":"class LandmarkEmbedding(tf.keras.Model):\n    def __init__(self, units, name):\n        super(LandmarkEmbedding, self).__init__(name=f'{name}_embedding')\n        self.units = units\n        \n    def build(self, input_shape):\n        # Embedding for missing landmark in frame, initizlied with zeros\n        self.empty_embedding = self.add_weight(\n            name=f'{self.name}_empty_embedding',\n            shape=[self.units],\n            initializer=INIT_ZEROS,\n        )\n        # Embedding\n        self.dense = tf.keras.Sequential([\n            tf.keras.layers.Dense(self.units, name=f'{self.name}_dense_1', use_bias=False, kernel_initializer=INIT_GLOROT_UNIFORM, activation=GELU),\n            tf.keras.layers.Dense(self.units, name=f'{self.name}_dense_2', use_bias=False, kernel_initializer=INIT_HE_UNIFORM),\n        ], name=f'{self.name}_dense')\n\n    def call(self, x):\n        return tf.where(\n                # Checks whether landmark is missing in frame\n                tf.reduce_sum(x, axis=2, keepdims=True) == 0,\n                # If so, the empty embedding is used\n                self.empty_embedding,\n                # Otherwise the landmark data is embedded\n                self.dense(x),\n            )","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:05:10.310404Z","iopub.execute_input":"2023-05-06T11:05:10.311171Z","iopub.status.idle":"2023-05-06T11:05:10.329288Z","shell.execute_reply.started":"2023-05-06T11:05:10.311119Z","shell.execute_reply":"2023-05-06T11:05:10.328209Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Embedding","metadata":{}},{"cell_type":"code","source":"class CustomEmbedding(tf.keras.Model):\n    def __init__(self):\n        super(CustomEmbedding, self).__init__()\n\n    def build(self, input_shape):\n        # Positional Embedding, initialized with zeros\n        self.positional_embedding = tf.keras.layers.Embedding(cfg.INPUT_SIZE+1, UNITS, embeddings_initializer=INIT_ZEROS)\n        # Embedding layer for Landmarks\n        self.lips_embedding = LandmarkEmbedding(LIPS_UNITS, 'lips')\n        self.left_hand_embedding = LandmarkEmbedding(HANDS_UNITS, 'left_hand')\n        self.right_hand_embedding = LandmarkEmbedding(HANDS_UNITS, 'right_hand')\n        self.pose_embedding = LandmarkEmbedding(POSE_UNITS, 'pose')\n        # Landmark Weights\n        self.landmark_weights = tf.Variable(tf.zeros([4], dtype=tf.float32), name='landmark_weights')\n        # Fully Connected Layers for combined landmarks\n        self.fc = tf.keras.Sequential([\n            tf.keras.layers.Dense(UNITS, name='fully_connected_1', use_bias=False, kernel_initializer=INIT_GLOROT_UNIFORM, activation=GELU),\n            tf.keras.layers.Dense(UNITS, name='fully_connected_2', use_bias=False, kernel_initializer=INIT_HE_UNIFORM),\n        ], name='fc')\n\n\n    def call(self, lips0, left_hand0, right_hand0, pose0, non_empty_frame_idxs, training=False):\n        # Lips\n        lips_embedding = self.lips_embedding(lips0)\n        # Left Hand\n        left_hand_embedding = self.left_hand_embedding(left_hand0)\n        # Right Hand\n        right_hand_embedding = self.right_hand_embedding(right_hand0)\n        # Pose\n        pose_embedding = self.pose_embedding(pose0)\n        # Merge Embeddings of all landmarks with mean pooling\n        x = tf.stack((lips_embedding, left_hand_embedding, right_hand_embedding, pose_embedding), axis=3)\n        # Merge Landmarks with trainable attention weights\n        x = x * tf.nn.softmax(self.landmark_weights)\n        x = tf.reduce_sum(x, axis=3)\n        # Fully Connected Layers\n        x = self.fc(x)\n        # Add Positional Embedding\n        normalised_non_empty_frame_idxs = tf.where(\n            tf.math.equal(non_empty_frame_idxs, -1.0),\n            cfg.INPUT_SIZE,\n            tf.cast(\n                non_empty_frame_idxs / tf.reduce_max(non_empty_frame_idxs, axis=1, keepdims=True) * cfg.INPUT_SIZE,\n                tf.int32,\n            ),\n        )\n        x = x + self.positional_embedding(normalised_non_empty_frame_idxs)\n        \n        return x","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:05:10.336476Z","iopub.execute_input":"2023-05-06T11:05:10.337136Z","iopub.status.idle":"2023-05-06T11:05:10.365198Z","shell.execute_reply.started":"2023-05-06T11:05:10.337091Z","shell.execute_reply":"2023-05-06T11:05:10.361628Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def get_lr_metric(optimizer):\n    def lr(y_true, y_pred):\n        return optimizer.lr\n    return lr","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:05:10.366665Z","iopub.execute_input":"2023-05-06T11:05:10.367310Z","iopub.status.idle":"2023-05-06T11:05:10.379701Z","shell.execute_reply.started":"2023-05-06T11:05:10.367271Z","shell.execute_reply":"2023-05-06T11:05:10.378740Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def get_model():\n    # Inputs\n    frames = tf.keras.layers.Input([cfg.INPUT_SIZE, N_COLS, cfg.N_DIMS], dtype=tf.float32, name='frames')\n    non_empty_frame_idxs = tf.keras.layers.Input([cfg.INPUT_SIZE], dtype=tf.float32, name='non_empty_frame_idxs')\n    # Padding Mask\n    mask = tf.cast(tf.math.not_equal(non_empty_frame_idxs, -1), tf.float32)\n    mask = tf.expand_dims(mask, axis=2)\n    \n    x = frames\n    x = tf.slice(x, [0,0,0,0], [-1,cfg.INPUT_SIZE, N_COLS, 2])\n    # LIPS\n    lips = tf.slice(x, [0,0,LIPS_START,0], [-1,cfg.INPUT_SIZE, 40, 2])\n    lips = tf.where(\n            tf.math.equal(lips, 0.0),\n            0.0,\n            (lips - LIPS_MEAN) / LIPS_STD,\n        )\n    lips = tf.reshape(lips, [-1, cfg.INPUT_SIZE, 40*2])\n    # LEFT HAND\n    left_hand = tf.slice(x, [0,0,40,0], [-1,cfg.INPUT_SIZE, 21, 2])\n    left_hand = tf.where(\n            tf.math.equal(left_hand, 0.0),\n            0.0,\n            (left_hand - LEFT_HANDS_MEAN) / LEFT_HANDS_STD,\n        )\n    left_hand = tf.reshape(left_hand, [-1, cfg.INPUT_SIZE, 21*2])\n    # RIGHT HAND\n    right_hand = tf.slice(x, [0,0,61,0], [-1,cfg.INPUT_SIZE, 21, 2])\n    right_hand = tf.where(\n            tf.math.equal(right_hand, 0.0),\n            0.0,\n            (right_hand - RIGHT_HANDS_MEAN) / RIGHT_HANDS_STD,\n        )\n    right_hand = tf.reshape(right_hand, [-1, cfg.INPUT_SIZE, 21*2])\n    # POSE\n    pose = tf.slice(x, [0,0,82,0], [-1,cfg.INPUT_SIZE, 10, 2])\n    pose = tf.where(\n            tf.math.equal(pose, 0.0),\n            0.0,\n            (pose - POSE_MEAN) / POSE_STD,\n        )\n    pose = tf.reshape(pose, [-1, cfg.INPUT_SIZE, 10*2])\n    x = lips, left_hand, right_hand, pose\n    x = CustomEmbedding()(lips, left_hand, right_hand, pose, non_empty_frame_idxs)\n    # Encoder Transformer Blocks\n    x = Transformer(NUM_BLOCKS)(x, mask)\n    # Pooling\n    x = tf.reduce_sum(x * mask, axis=1) / tf.reduce_sum(mask, axis=1)\n    # Classification Layer\n    x = tf.keras.layers.Dense(cfg.NUM_CLASSES, activation=tf.keras.activations.softmax, kernel_initializer=INIT_GLOROT_UNIFORM)(x)\n    outputs = x\n    \n    # Create Tensorflow Model\n    model = tf.keras.models.Model(inputs=[frames, non_empty_frame_idxs], outputs=outputs)\n    \n    # Simple Categorical Crossentropy Loss\n    loss = tf.keras.losses.SparseCategoricalCrossentropy()\n    \n    # Adam Optimizer with weight decay\n    optimizer = tfa.optimizers.AdamW(learning_rate=1e-3, weight_decay=1e-5, clipnorm=1.0)\n    \n    lr_metric = get_lr_metric(optimizer)\n    metrics = [\"acc\",lr_metric]\n    model.compile(loss=loss, optimizer=optimizer, metrics=metrics)\n    return model","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:05:10.381003Z","iopub.execute_input":"2023-05-06T11:05:10.381531Z","iopub.status.idle":"2023-05-06T11:05:10.405099Z","shell.execute_reply.started":"2023-05-06T11:05:10.381493Z","shell.execute_reply":"2023-05-06T11:05:10.404125Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"tf.keras.backend.clear_session()\n\nmodel_one = get_model()","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:05:10.408990Z","iopub.execute_input":"2023-05-06T11:05:10.409646Z","iopub.status.idle":"2023-05-06T11:05:13.435890Z","shell.execute_reply.started":"2023-05-06T11:05:10.409610Z","shell.execute_reply":"2023-05-06T11:05:13.434886Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"model_one.summary()","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:07:19.169089Z","iopub.execute_input":"2023-05-06T11:07:19.169503Z","iopub.status.idle":"2023-05-06T11:07:19.260531Z","shell.execute_reply.started":"2023-05-06T11:07:19.169465Z","shell.execute_reply":"2023-05-06T11:07:19.259739Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def lrfn(current_step, num_warmup_steps, lr_max, num_cycles=0.50, num_training_steps=cfg.N_EPOCHS):\n    \n    if current_step < num_warmup_steps:\n        if WARMUP_METHOD == 'log':\n            return lr_max * 0.10 ** (num_warmup_steps - current_step)\n        else:\n            return lr_max * 2 ** -(num_warmup_steps - current_step)\n    else:\n        progress = float(current_step - num_warmup_steps) / float(max(1, num_training_steps - num_warmup_steps))\n\n        return max(0.0, 0.5 * (1.0 + math.cos(math.pi * float(num_cycles) * 2.0 * progress))) * lr_max","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:07:24.418350Z","iopub.execute_input":"2023-05-06T11:07:24.418773Z","iopub.status.idle":"2023-05-06T11:07:24.425763Z","shell.execute_reply.started":"2023-05-06T11:07:24.418738Z","shell.execute_reply":"2023-05-06T11:07:24.424340Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def plot_lr_schedule(lr_schedule, epochs):\n    fig = plt.figure(figsize=(20, 10))\n    plt.plot([None] + lr_schedule + [None])\n    # X Labels\n    x = np.arange(1, epochs + 1)\n    x_axis_labels = [i if epochs <= 40 or i % 5 == 0 or i == 1 else None for i in range(1, epochs + 1)]\n    plt.xlim([1, epochs])\n    plt.xticks(x, x_axis_labels) # set tick step to 1 and let x axis start at 1\n    \n    # Increase y-limit for better readability\n    plt.ylim([0, max(lr_schedule) * 1.1])\n    \n    # Title\n    schedule_info = f'start: {lr_schedule[0]:.1E}, max: {max(lr_schedule):.1E}, final: {lr_schedule[-1]:.1E}'\n    plt.title(f'Step Learning Rate Schedule, {schedule_info}', size=18, pad=12)\n    \n    # Plot Learning Rates\n    for x, val in enumerate(lr_schedule):\n        if epochs <= 40 or x % 5 == 0 or x is epochs - 1:\n            if x < len(lr_schedule) - 1:\n                if lr_schedule[x - 1] < val:\n                    ha = 'right'\n                else:\n                    ha = 'left'\n            elif x == 0:\n                ha = 'right'\n            else:\n                ha = 'left'\n            plt.plot(x + 1, val, 'o', color='black');\n            offset_y = (max(lr_schedule) - min(lr_schedule)) * 0.02\n            plt.annotate(f'{val:.1E}', xy=(x + 1, val + offset_y), size=12, ha=ha)\n    \n    plt.xlabel('Epoch', size=16, labelpad=5)\n    plt.ylabel('Learning Rate', size=16, labelpad=5)\n    plt.grid()\n    plt.show()\n\n# Learning rate for encoder\nLR_SCHEDULE = [lrfn(step, num_warmup_steps=cfg.N_WARMUP_EPOCHS, lr_max=cfg.LR_MAX, num_cycles=0.50) for step in range(cfg.N_EPOCHS)]\n# Plot Learning Rate Schedule\nplot_lr_schedule(LR_SCHEDULE, epochs=cfg.N_EPOCHS)\n# Learning Rate Callback\nlr_callback = tf.keras.callbacks.LearningRateScheduler(lambda step: LR_SCHEDULE[step], verbose=1)","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:07:26.319523Z","iopub.execute_input":"2023-05-06T11:07:26.320010Z","iopub.status.idle":"2023-05-06T11:07:26.842071Z","shell.execute_reply.started":"2023-05-06T11:07:26.319972Z","shell.execute_reply":"2023-05-06T11:07:26.841001Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Weight Decay Callback","metadata":{}},{"cell_type":"code","source":"class WeightDecayCallback(tf.keras.callbacks.Callback):\n    def __init__(self, wd_ratio=cfg.WD_RATIO):\n        self.step_counter = 0\n        self.wd_ratio = wd_ratio\n    \n    def on_epoch_begin(self, epoch, logs=None):\n        model_one.optimizer.weight_decay = model_one.optimizer.learning_rate * self.wd_ratio\n        print(f'learning rate: {model_one.optimizer.learning_rate.numpy():.2e}, weight decay: {model_one.optimizer.weight_decay.numpy():.2e}')","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:07:31.209765Z","iopub.execute_input":"2023-05-06T11:07:31.210750Z","iopub.status.idle":"2023-05-06T11:07:31.218736Z","shell.execute_reply.started":"2023-05-06T11:07:31.210711Z","shell.execute_reply":"2023-05-06T11:07:31.217641Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Performance Benchmark","metadata":{}},{"cell_type":"code","source":"%%timeit -n 100\nif cfg.TRAIN_MODEL:\n    # Verify model_one prediction is <<<100ms\n    model_one.predict_on_batch({ 'frames': X[:1], 'non_empty_frame_idxs': NON_EMPTY_FRAME_IDXS[:1] })","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:07:33.453415Z","iopub.execute_input":"2023-05-06T11:07:33.454446Z","iopub.status.idle":"2023-05-06T11:07:43.284155Z","shell.execute_reply.started":"2023-05-06T11:07:33.454392Z","shell.execute_reply":"2023-05-06T11:07:43.282972Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"## Training","metadata":{}},{"cell_type":"code","source":"X_train = X[train_idxs]\nX_val = X[val_idxs]\nNON_EMPTY_FRAME_IDXS_TRAIN = NON_EMPTY_FRAME_IDXS[train_idxs]\nNON_EMPTY_FRAME_IDXS_VAL = NON_EMPTY_FRAME_IDXS[val_idxs]\ny_train = y[train_idxs]\ny_val = y[val_idxs]","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:08:02.213649Z","iopub.execute_input":"2023-05-06T11:08:02.214504Z","iopub.status.idle":"2023-05-06T11:08:05.939210Z","shell.execute_reply.started":"2023-05-06T11:08:02.214454Z","shell.execute_reply":"2023-05-06T11:08:05.937968Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"if cfg.TRAIN_MODEL:\n    run = wandb.init(project=\"kaggle-asl-signs\", config=cfg, tags=['transformer', 'final-model'])","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:08:07.447029Z","iopub.execute_input":"2023-05-06T11:08:07.447505Z","iopub.status.idle":"2023-05-06T11:08:38.988736Z","shell.execute_reply.started":"2023-05-06T11:08:07.447459Z","shell.execute_reply":"2023-05-06T11:08:38.987679Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"if cfg.TRAIN_MODEL:\n    tf.keras.backend.clear_session()\n    callbacks=[\n            lr_callback,\n            WeightDecayCallback(),\n            wandb.keras.WandbCallback()\n    ]\n    model_one.fit(\n        x=get_train_batch_all_signs(X, y, NON_EMPTY_FRAME_IDXS),\n        steps_per_epoch=len(X) // (cfg.NUM_CLASSES * cfg.BATCH_ALL_SIGNS_N),\n        epochs=cfg.N_EPOCHS,\n        batch_size=cfg.BATCH_SIZE,\n        callbacks=callbacks,\n        verbose = 2,) ","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:10:08.081567Z","iopub.execute_input":"2023-05-06T11:10:08.082291Z","iopub.status.idle":"2023-05-06T11:31:49.717108Z","shell.execute_reply.started":"2023-05-06T11:10:08.082249Z","shell.execute_reply":"2023-05-06T11:31:49.715938Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"if cfg.TRAIN_MODEL:\n    model_one.save('./final_model_one')\n    model_one.save_weights('./final_model_one_weights')","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:43:11.858834Z","iopub.execute_input":"2023-05-06T11:43:11.859511Z","iopub.status.idle":"2023-05-06T11:43:34.440436Z","shell.execute_reply.started":"2023-05-06T11:43:11.859472Z","shell.execute_reply":"2023-05-06T11:43:34.439185Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"if cfg.TRAIN_MODEL:\n    artifact = wandb.Artifact('final_model_one', type='model')\n    artifact.add_file('./final_model_one_weights.data-00000-of-00001')\n    artifact.add_file('./final_model_one_weights.index')\n    run.log_artifact(artifact)","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:44:22.235356Z","iopub.execute_input":"2023-05-06T11:44:22.236031Z","iopub.status.idle":"2023-05-06T11:44:22.566795Z","shell.execute_reply.started":"2023-05-06T11:44:22.235988Z","shell.execute_reply":"2023-05-06T11:44:22.565653Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# download model\ntry:\n    artifact = run.use_artifact('amanarora/asl-sings/final_model_one:v1', type='model')\nexcept NameError:\n    run = wandb.init(project=\"kaggle-asl-signs\")\n    artifact = run.use_artifact('amanarora/asl-sings/final_model_one:v1', type='model')\nartifact_dir = artifact.download()\n!ls {artifact_dir}","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:46:13.874570Z","iopub.execute_input":"2023-05-06T11:46:13.875532Z","iopub.status.idle":"2023-05-06T11:46:16.997776Z","shell.execute_reply.started":"2023-05-06T11:46:13.875481Z","shell.execute_reply":"2023-05-06T11:46:16.996475Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# load model weights and do sanity c\nmodel_one.load_weights(f\"{artifact_dir}/final_model_one_weights\")\ny_val_pred = model_one.predict({ 'frames': X_val, 'non_empty_frame_idxs': NON_EMPTY_FRAME_IDXS_VAL }, verbose=2).argmax(axis=1)\nnp.mean(y_val == y_val_pred)","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:47:07.031067Z","iopub.execute_input":"2023-05-06T11:47:07.031473Z","iopub.status.idle":"2023-05-06T11:47:15.497579Z","shell.execute_reply.started":"2023-05-06T11:47:07.031433Z","shell.execute_reply":"2023-05-06T11:47:15.495617Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# delete variables as we go to free up RAM\ndel X; del y; del X_train; del X_val; del y_train; del y_val","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:47:44.741650Z","iopub.execute_input":"2023-05-06T11:47:44.742679Z","iopub.status.idle":"2023-05-06T11:47:44.852197Z","shell.execute_reply.started":"2023-05-06T11:47:44.742634Z","shell.execute_reply":"2023-05-06T11:47:44.850918Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Landmark Weights\nweights = scipy.special.softmax(model_one.get_layer('custom_embedding').weights[15])\nlandmarks = ['lips_embedding', 'left_hand_embedding', 'right_hand_embedding', 'pose_embedding']\n\n# Learned attention weights, initialized at uniform 25%\nfor w, lm in zip(weights, landmarks):\n    print(f'{lm} weight: {(w*100):.1f}%')","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:48:01.386281Z","iopub.execute_input":"2023-05-06T11:48:01.387105Z","iopub.status.idle":"2023-05-06T11:48:01.411592Z","shell.execute_reply.started":"2023-05-06T11:48:01.387063Z","shell.execute_reply":"2023-05-06T11:48:01.407474Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"converter = tf.lite.TFLiteConverter.from_keras_model(model_one)\nconverter.optimizations = [tf.lite.Optimize.DEFAULT]\ntflite_model = converter.convert()\nwith open('./model.tflite', 'wb') as f:\n    f.write(tflite_model)","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:54:19.221140Z","iopub.execute_input":"2023-05-06T11:54:19.221560Z","iopub.status.idle":"2023-05-06T11:55:01.465057Z","shell.execute_reply.started":"2023-05-06T11:54:19.221522Z","shell.execute_reply":"2023-05-06T11:55:01.463802Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"!du -sh ./model.tflite","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:55:24.132336Z","iopub.execute_input":"2023-05-06T11:55:24.132716Z","iopub.status.idle":"2023-05-06T11:55:25.367415Z","shell.execute_reply.started":"2023-05-06T11:55:24.132677Z","shell.execute_reply":"2023-05-06T11:55:25.366194Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"!zip submission.zip ./model.tflite","metadata":{"execution":{"iopub.status.busy":"2023-05-06T11:55:40.587454Z","iopub.execute_input":"2023-05-06T11:55:40.588483Z","iopub.status.idle":"2023-05-06T11:55:42.090855Z","shell.execute_reply.started":"2023-05-06T11:55:40.588431Z","shell.execute_reply":"2023-05-06T11:55:42.089421Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]}]}