Sign2TextConverter.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 156,
   "metadata": {},
   "outputs": [],
   "source": [
    "import mediapipe as mp\n",
    "import cv2\n",
    "import numpy as np\n",
    "import os\n",
    "import uuid"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 157,
   "metadata": {},
   "outputs": [],
   "source": [
    "mp_holistic = mp.solutions.holistic # Holistic model\n",
    "mp_drawing = mp.solutions.drawing_utils # Drawing utilities\n",
    "\n",
    "def mediapipe_detection(image, model):\n",
    "    image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB) # COLOR CONVERSION BGR 2 RGB\n",
    "    image.flags.writeable = False # Image is no longer writeable\n",
    "    results = model.process(image) # Make prediction\n",
    "    image.flags.writeable = True # Image is now writeable \n",
    "    image = cv2.cvtColor(image, cv2.COLOR_RGB2BGR) # COLOR COVERSION RGB 2 BGR\n",
    "    return image, results\n",
    "\n",
    "def draw_landmarks(image, results):\n",
    "    mp_drawing.draw_landmarks(image, results.face_landmarks, mp_holistic.FACEMESH_CONTOURS) # Draw face connections\n",
    "    mp_drawing.draw_landmarks(image, results.pose_landmarks, mp_holistic.POSE_CONNECTIONS) # Draw pose connections\n",
    "    mp_drawing.draw_landmarks(image, results.left_hand_landmarks, mp_holistic.HAND_CONNECTIONS) # Draw left hand connections\n",
    "    mp_drawing.draw_landmarks(image, results.right_hand_landmarks, mp_holistic.HAND_CONNECTIONS) # Draw right hand connections\n",
    "    \n",
    "def draw_styled_landmarks(image, results):\n",
    "    # Draw face connections\n",
    "    mp_drawing.draw_landmarks(image, results.face_landmarks, mp_holistic.FACEMESH_CONTOURS, \n",
    "                             mp_drawing.DrawingSpec(color=(80,110,10), thickness=1, circle_radius=1), \n",
    "                             mp_drawing.DrawingSpec(color=(80,256,121), thickness=1, circle_radius=1)\n",
    "                             ) \n",
    "    # Draw pose connections\n",
    "    mp_drawing.draw_landmarks(image, results.pose_landmarks, mp_holistic.POSE_CONNECTIONS,\n",
    "                             mp_drawing.DrawingSpec(color=(80,22,10), thickness=2, circle_radius=4), \n",
    "                             mp_drawing.DrawingSpec(color=(80,44,121), thickness=2, circle_radius=2)\n",
    "                             ) \n",
    "    # Draw left hand connections\n",
    "    mp_drawing.draw_landmarks(image, results.left_hand_landmarks, mp_holistic.HAND_CONNECTIONS, \n",
    "                             mp_drawing.DrawingSpec(color=(121,22,76), thickness=2, circle_radius=4), \n",
    "                             mp_drawing.DrawingSpec(color=(121,44,250), thickness=2, circle_radius=2)\n",
    "                             ) \n",
    "    # Draw right hand connections  \n",
    "    mp_drawing.draw_landmarks(image, results.right_hand_landmarks, mp_holistic.HAND_CONNECTIONS, \n",
    "                             mp_drawing.DrawingSpec(color=(245,117,66), thickness=2, circle_radius=4), \n",
    "                             mp_drawing.DrawingSpec(color=(245,66,230), thickness=2, circle_radius=2)\n",
    "                             )\n",
    "    \n",
    "    \n",
    "def extract_keypoints(results):\n",
    "    pose = np.array([[res.x, res.y, res.z, res.visibility] for res in results.pose_landmarks.landmark]).flatten() if results.pose_landmarks else np.zeros(33*4)\n",
    "    \n",
    "    face = np.array([[res.x, res.y, res.z] for res in results.face_landmarks.landmark]).flatten() if results.face_landmarks else np.zeros(468*3)\n",
    "    \n",
    "    lh = np.array([[res.x, res.y, res.z] for res in results.left_hand_landmarks.landmark]).flatten() if results.left_hand_landmarks else np.zeros(21*3)\n",
    "    \n",
    "    rh = np.array([[res.x, res.y, res.z] for res in results.right_hand_landmarks.landmark]).flatten() if results.right_hand_landmarks else np.zeros(21*3)\n",
    "    \n",
    "    # return np.concatenate([lh, rh])\n",
    "    return np.concatenate([pose, face, lh, rh])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 161,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Path for exported data, numpy arrays\n",
    "\n",
    "# DATA_PATH = os.path.join(\"Dataset\")\n",
    "# DATA_PATH = os.path.join(\"Dataset2\")\n",
    "DATA_PATH = os.path.join(\"Landmarks-Dataset\")\n",
    "\n",
    "# Actions that we try to detect\n",
    "# actions = np.array(['Yellow'])\n",
    "# actions = np.array(['White','Gray','Black','Goodbye','Hello'])\n",
    "actions = np.array(['a','aa','au'])\n",
    "\n",
    "# Thirty videos worth of data\n",
    "no_sequences = 30\n",
    "\n",
    "# Videos are going to be 30 frames in length\n",
    "# sequence_length = frame_count\n",
    "\n",
    "# Folder start\n",
    "start_folder = 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 162,
   "metadata": {},
   "outputs": [],
   "source": [
    "for action in actions:\n",
    "    # dirmax = np.max(np.array(os.listdir(os.path.join(DATA_PATH, action))).astype(int))\n",
    "    for sequence in range(1, no_sequences+1):\n",
    "        try:\n",
    "            os.makedirs(os.path.join(DATA_PATH, action, str(sequence)))\n",
    "            # os.makedirs(os.path.join(DATA_PATH, action, str(dirmax + sequence)))\n",
    "        except:\n",
    "            pass"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 163,
   "metadata": {},
   "outputs": [],
   "source": [
    "try:\n",
    "    # cap = cv2.VideoCapture('./Dataset/{}/{} - {}.mp4'.format(action,action,sequence))\n",
    "    # frame_count = cap.get(cv2.CAP_PROP_FRAME_COUNT)\n",
    "    # # Thirty videos worth of data\n",
    "    # sequence_length = frame_count\n",
    "    # Set mediapipe model \n",
    "    with mp_holistic.Holistic(min_detection_confidence=0.5, min_tracking_confidence=0.5) as holistic:\n",
    "        \n",
    "        # NEW LOOP\n",
    "        # Loop through actions\n",
    "        for action in actions:\n",
    "        #     # Loop through sequences aka videos\n",
    "            for sequence in range(1,no_sequences+1):\n",
    "                cap = cv2.VideoCapture('./Dataset/{}/{} - {}.mp4'.format(action,action,sequence))\n",
    "                frame_count = cap.get(cv2.CAP_PROP_FRAME_COUNT)\n",
    "                # Thirty videos worth of data\n",
    "                # sequence_length = frame_count\n",
    "        #         # Loop through video length aka sequence length\n",
    "                for frame_num in range(int(frame_count)):\n",
    "                # for frame_num in range(50):\n",
    "\n",
    "        # Read feed\n",
    "                    # cap = cv2.VideoCapture('./Dataset/{}/{} - {}.mp4'.format(action,action,sequence))\n",
    "                    ret, frame = cap.read()\n",
    "\n",
    "                    # Make detections\n",
    "                    image, results = mediapipe_detection(frame, holistic)\n",
    "\n",
    "                    # Draw landmarks\n",
    "                    draw_styled_landmarks(image, results)\n",
    "                    \n",
    "                    # NEW Apply wait logic\n",
    "                    if frame_num == 1: \n",
    "                    #     cv2.putText(image, 'STARTING COLLECTION', (120,200), \n",
    "                    #             cv2.FONT_HERSHEY_SIMPLEX, 1, (0,255, 0), 4, cv2.LINE_AA)\n",
    "                    #     cv2.putText(image, 'Collecting frames for {} Video Number {}'.format(action, sequence), (15,12), \n",
    "                    #             cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 0, 255), 1, cv2.LINE_AA)\n",
    "                    #     # Show to screen\n",
    "                        cv2.imshow('OpenCV Feed', image)\n",
    "                        # cv2.waitKey(1500)\n",
    "                    else: \n",
    "                    #     cv2.putText(image, 'Collecting frames for {} Video Number {}'.format(action, sequence), (15,12), \n",
    "                    #             cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 0, 255), 1, cv2.LINE_AA)\n",
    "                    #     # Show to screen\n",
    "                        cv2.imshow('OpenCV Feed', image)\n",
    "                    \n",
    "                    # NEW Export keypoints\n",
    "                    keypoints = extract_keypoints(results)\n",
    "                    npy_path = os.path.join(DATA_PATH, action, str(sequence), str(frame_num))\n",
    "                    np.save(npy_path, keypoints)\n",
    "\n",
    "                                # Break gracefully\n",
    "                    if cv2.waitKey(10) & 0xFF == ord('q'):\n",
    "                        break\n",
    "                        \n",
    "        cap.release()\n",
    "        cv2.destroyAllWindows()\n",
    "        \n",
    "finally:\n",
    "    cap.release()\n",
    "    cv2.destroyAllWindows()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 164,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "from tensorflow.keras.utils import to_categorical"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 165,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'a': 0, 'aa': 1, 'au': 2}\n"
     ]
    }
   ],
   "source": [
    "label_map = {label: num for num, label in enumerate(actions)}\n",
    "print(label_map)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 166,
   "metadata": {},
   "outputs": [],
   "source": [
    "dir_path=DATA_PATH"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 167,
   "metadata": {},
   "outputs": [],
   "source": [
    "# sequences, labels = [], []\n",
    "# for action in actions:\n",
    "#     for sequence in np.array(os.listdir(os.path.join(DATA_PATH, action))).astype(int):\n",
    "#         window = []\n",
    "#         for frame_num in range(1,sequence_length+1):\n",
    "#             res = np.load(os.path.join(DATA_PATH, action, str(sequence), \"{}.npy\".format(frame_num)))\n",
    "#             window.append(res)\n",
    "#         sequences.append(window)\n",
    "#         labels.append(label_map[action])\n",
    "\n",
    "# sequences, labels = [], []\n",
    "# for action in actions:\n",
    "#     for sequence in np.array(os.listdir(os.path.join(DATA_PATH, action))).astype(int):\n",
    "#         window = []\n",
    "#         num_frames = len(os.listdir(os.path.join(DATA_PATH, action, str(sequence))))\n",
    "#         for frame_num in range(num_frames):\n",
    "#             res = np.load(os.path.join(DATA_PATH, action, str(sequence), \"{}.npy\".format(frame_num)))\n",
    "#             window.append(res)\n",
    "#         sequences.append(window)\n",
    "#         labels.append(label_map[action])\n",
    "\n",
    "sequences, labels = [], []\n",
    "max_length = 0\n",
    "for action in actions:\n",
    "    for sequence in np.array(os.listdir(os.path.join(DATA_PATH, action))).astype(int):\n",
    "        window = []\n",
    "        num_frames = len(os.listdir(os.path.join(DATA_PATH, action, str(sequence))))\n",
    "        if num_frames > max_length:\n",
    "            max_length = num_frames\n",
    "        for frame_num in range(num_frames):\n",
    "            res = np.load(os.path.join(DATA_PATH, action, str(sequence), \"{}.npy\".format(frame_num)))\n",
    "            window.append(res)\n",
    "        sequences.append(window)\n",
    "        labels.append(label_map[action])\n",
    "\n",
    "# Pad sequences to maximum length\n",
    "for i, seq in enumerate(sequences):\n",
    "    seq_length = len(seq)\n",
    "    if seq_length < max_length:\n",
    "        num_padding_frames = max_length - seq_length\n",
    "        padding_frames = [np.zeros_like(seq[0]) for _ in range(num_padding_frames)]\n",
    "        sequences[i] = seq + padding_frames\n",
    "\n",
    "sequences = np.array(sequences)\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 168,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "90"
      ]
     },
     "execution_count": 168,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(sequences)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 169,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(90, 82, 1662)"
      ]
     },
     "execution_count": 169,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.array(sequences).shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 170,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(90,)"
      ]
     },
     "execution_count": 170,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.array(labels).shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 171,
   "metadata": {},
   "outputs": [],
   "source": [
    "x = np.asarray(sequences)\n",
    "y = to_categorical(labels).astype(int)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 172,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(90, 82, 1662)"
      ]
     },
     "execution_count": 172,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 173,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [1, 0, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 1, 0],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1],\n",
       "       [0, 0, 1]])"
      ]
     },
     "execution_count": 173,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 174,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.05)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 175,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(85, 82, 1662)"
      ]
     },
     "execution_count": 175,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 176,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(5, 3)"
      ]
     },
     "execution_count": 176,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 177,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import LSTM, Dense\n",
    "from tensorflow.keras.callbacks import TensorBoard"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 178,
   "metadata": {},
   "outputs": [],
   "source": [
    "log_dir = os.path.join('Logs')\n",
    "# tb_callback = TensorBoard(log_dir=log_dir)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 179,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['a', 'aa', 'au'], dtype='<U2')"
      ]
     },
     "execution_count": 179,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "actions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 186,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = Sequential()\n",
    "model.add(LSTM(64, return_sequences=True, activation='relu', input_shape=(82,1662)))\n",
    "model.add(LSTM(128, return_sequences=True, activation='relu'))\n",
    "model.add(LSTM(64, return_sequences=False, activation='relu'))\n",
    "model.add(Dense(64, activation='relu'))\n",
    "model.add(Dense(32, activation='relu'))\n",
    "model.add(Dense(actions.shape[0], activation='softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 187,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "You must install pydot (`pip install pydot`) and install graphviz (see instructions at https://graphviz.gitlab.io/download/) for plot_model/model_to_dot to work.\n"
     ]
    }
   ],
   "source": [
    "from keras.utils import plot_model\n",
    "plot_model(model, to_file='model.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 188,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['a', 'aa', 'au'], dtype='<U2')"
      ]
     },
     "execution_count": 188,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "actions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 189,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(90, 82, 1662)"
      ]
     },
     "execution_count": 189,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 190,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.compile(optimizer='Adam',loss='categorical_crossentropy',metrics=['categorical_accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 191,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/200\n",
      "3/3 [==============================] - 3s 201ms/step - loss: 341.3217 - categorical_accuracy: 0.4353\n",
      "Epoch 2/200\n",
      "3/3 [==============================] - 1s 199ms/step - loss: 51.4110 - categorical_accuracy: 0.3294\n",
      "Epoch 3/200\n",
      "3/3 [==============================] - 1s 185ms/step - loss: 121.1545 - categorical_accuracy: 0.3176\n",
      "Epoch 4/200\n",
      "3/3 [==============================] - 1s 189ms/step - loss: 536.5491 - categorical_accuracy: 0.3294\n",
      "Epoch 5/200\n",
      "3/3 [==============================] - 1s 188ms/step - loss: 1801.7623 - categorical_accuracy: 0.3412\n",
      "Epoch 6/200\n",
      "3/3 [==============================] - 1s 189ms/step - loss: 17461.7090 - categorical_accuracy: 0.3412\n",
      "Epoch 7/200\n",
      "3/3 [==============================] - 1s 201ms/step - loss: 15912.8174 - categorical_accuracy: 0.4000\n",
      "Epoch 8/200\n",
      "3/3 [==============================] - 1s 209ms/step - loss: 37425.4414 - categorical_accuracy: 0.3765\n",
      "Epoch 9/200\n",
      "3/3 [==============================] - 1s 199ms/step - loss: 27179.7539 - categorical_accuracy: 0.3765\n",
      "Epoch 10/200\n",
      "3/3 [==============================] - 1s 184ms/step - loss: 77743.2500 - categorical_accuracy: 0.2588\n",
      "Epoch 11/200\n",
      "3/3 [==============================] - 1s 209ms/step - loss: 36548.9336 - categorical_accuracy: 0.3294\n",
      "Epoch 12/200\n",
      "3/3 [==============================] - 1s 215ms/step - loss: 51079.7773 - categorical_accuracy: 0.2471\n",
      "Epoch 13/200\n",
      "3/3 [==============================] - 1s 212ms/step - loss: 97750.4375 - categorical_accuracy: 0.2824\n",
      "Epoch 14/200\n",
      "3/3 [==============================] - 1s 179ms/step - loss: 92494.3672 - categorical_accuracy: 0.3059\n",
      "Epoch 15/200\n",
      "3/3 [==============================] - 1s 208ms/step - loss: 127944.6016 - categorical_accuracy: 0.4000\n",
      "Epoch 16/200\n",
      "3/3 [==============================] - 1s 184ms/step - loss: 46127.2188 - categorical_accuracy: 0.4353\n",
      "Epoch 17/200\n",
      "3/3 [==============================] - 1s 199ms/step - loss: 34650.6289 - categorical_accuracy: 0.3529\n",
      "Epoch 18/200\n",
      "3/3 [==============================] - 1s 214ms/step - loss: 30988.0625 - categorical_accuracy: 0.3176\n",
      "Epoch 19/200\n",
      "3/3 [==============================] - 1s 194ms/step - loss: 29374.4238 - categorical_accuracy: 0.3529\n",
      "Epoch 20/200\n",
      "3/3 [==============================] - 1s 184ms/step - loss: 39639.3711 - categorical_accuracy: 0.3647\n",
      "Epoch 21/200\n",
      "3/3 [==============================] - 1s 192ms/step - loss: 8782.8555 - categorical_accuracy: 0.3294\n",
      "Epoch 22/200\n",
      "3/3 [==============================] - 1s 192ms/step - loss: 9334.9707 - categorical_accuracy: 0.3176\n",
      "Epoch 23/200\n",
      "3/3 [==============================] - 1s 183ms/step - loss: 7351.5161 - categorical_accuracy: 0.3294\n",
      "Epoch 24/200\n",
      "3/3 [==============================] - 1s 185ms/step - loss: 7882.6499 - categorical_accuracy: 0.3529\n",
      "Epoch 25/200\n",
      "3/3 [==============================] - 1s 208ms/step - loss: 11639.3545 - categorical_accuracy: 0.2824\n",
      "Epoch 26/200\n",
      "3/3 [==============================] - 1s 183ms/step - loss: 7812.6323 - categorical_accuracy: 0.3412\n",
      "Epoch 27/200\n",
      "3/3 [==============================] - 1s 184ms/step - loss: 9502.8145 - categorical_accuracy: 0.3765\n",
      "Epoch 28/200\n",
      "3/3 [==============================] - 1s 186ms/step - loss: 14886.2236 - categorical_accuracy: 0.3294\n",
      "Epoch 29/200\n",
      "3/3 [==============================] - 1s 209ms/step - loss: 11518.0430 - categorical_accuracy: 0.3765\n",
      "Epoch 30/200\n",
      "3/3 [==============================] - 1s 183ms/step - loss: 14091.9619 - categorical_accuracy: 0.2941\n",
      "Epoch 31/200\n",
      "3/3 [==============================] - 1s 190ms/step - loss: 13771.6768 - categorical_accuracy: 0.3647\n",
      "Epoch 32/200\n",
      "3/3 [==============================] - 1s 187ms/step - loss: 15430.1826 - categorical_accuracy: 0.3882\n",
      "Epoch 33/200\n",
      "3/3 [==============================] - 1s 183ms/step - loss: 43396.4883 - categorical_accuracy: 0.3412\n",
      "Epoch 34/200\n",
      "3/3 [==============================] - 1s 188ms/step - loss: 43053.1250 - categorical_accuracy: 0.3412\n",
      "Epoch 35/200\n",
      "3/3 [==============================] - 1s 188ms/step - loss: 98476.1562 - categorical_accuracy: 0.3529\n",
      "Epoch 36/200\n",
      "3/3 [==============================] - 1s 185ms/step - loss: 121623.9375 - categorical_accuracy: 0.3176\n",
      "Epoch 37/200\n",
      "3/3 [==============================] - 1s 187ms/step - loss: 679187.7500 - categorical_accuracy: 0.2941\n",
      "Epoch 38/200\n",
      "3/3 [==============================] - 1s 183ms/step - loss: 199475.8125 - categorical_accuracy: 0.2941\n",
      "Epoch 39/200\n",
      "3/3 [==============================] - 1s 188ms/step - loss: 248251.5312 - categorical_accuracy: 0.3294\n",
      "Epoch 40/200\n",
      "3/3 [==============================] - 1s 188ms/step - loss: 283055.0000 - categorical_accuracy: 0.3294\n",
      "Epoch 41/200\n",
      "3/3 [==============================] - 1s 187ms/step - loss: 227672.8438 - categorical_accuracy: 0.3294\n",
      "Epoch 42/200\n",
      "3/3 [==============================] - 1s 188ms/step - loss: 218272.7969 - categorical_accuracy: 0.3765\n",
      "Epoch 43/200\n",
      "3/3 [==============================] - 1s 210ms/step - loss: 183066.0781 - categorical_accuracy: 0.3412\n",
      "Epoch 44/200\n",
      "3/3 [==============================] - 1s 181ms/step - loss: 351680.1875 - categorical_accuracy: 0.2471\n",
      "Epoch 45/200\n",
      "3/3 [==============================] - 1s 185ms/step - loss: 145078.2344 - categorical_accuracy: 0.2941\n",
      "Epoch 46/200\n",
      "3/3 [==============================] - 1s 186ms/step - loss: 327236.9062 - categorical_accuracy: 0.2471\n",
      "Epoch 47/200\n",
      "3/3 [==============================] - 1s 187ms/step - loss: 501210.6875 - categorical_accuracy: 0.2471\n",
      "Epoch 48/200\n",
      "3/3 [==============================] - 1s 188ms/step - loss: 511328.9375 - categorical_accuracy: 0.3529\n",
      "Epoch 49/200\n",
      "3/3 [==============================] - 1s 190ms/step - loss: 292695.8125 - categorical_accuracy: 0.4471\n",
      "Epoch 50/200\n",
      "3/3 [==============================] - 1s 183ms/step - loss: 89509.5469 - categorical_accuracy: 0.4471\n",
      "Epoch 51/200\n",
      "3/3 [==============================] - 1s 189ms/step - loss: 329808.9062 - categorical_accuracy: 0.2235\n",
      "Epoch 52/200\n",
      "3/3 [==============================] - 1s 186ms/step - loss: 358079.0000 - categorical_accuracy: 0.3294\n",
      "Epoch 53/200\n",
      "3/3 [==============================] - 1s 188ms/step - loss: 58377.0000 - categorical_accuracy: 0.2824\n",
      "Epoch 54/200\n",
      "3/3 [==============================] - 1s 185ms/step - loss: 28540.4941 - categorical_accuracy: 0.3294\n",
      "Epoch 55/200\n",
      "3/3 [==============================] - 1s 188ms/step - loss: 33737.4648 - categorical_accuracy: 0.3529\n",
      "Epoch 56/200\n",
      "3/3 [==============================] - 1s 189ms/step - loss: 29438.6797 - categorical_accuracy: 0.2824\n",
      "Epoch 57/200\n",
      "3/3 [==============================] - 1s 183ms/step - loss: 16702.5332 - categorical_accuracy: 0.3412\n",
      "Epoch 58/200\n",
      "3/3 [==============================] - 1s 205ms/step - loss: 16889.6699 - categorical_accuracy: 0.3529\n",
      "Epoch 59/200\n",
      "3/3 [==============================] - 1s 184ms/step - loss: 23728.9883 - categorical_accuracy: 0.3294\n",
      "Epoch 60/200\n",
      "3/3 [==============================] - 1s 184ms/step - loss: 365860.5312 - categorical_accuracy: 0.3412\n",
      "Epoch 61/200\n",
      "3/3 [==============================] - 1s 187ms/step - loss: 99645.4844 - categorical_accuracy: 0.3647\n",
      "Epoch 62/200\n",
      "3/3 [==============================] - 1s 185ms/step - loss: 180566.8750 - categorical_accuracy: 0.3647\n",
      "Epoch 63/200\n",
      "3/3 [==============================] - 1s 187ms/step - loss: 57126.1602 - categorical_accuracy: 0.3529\n",
      "Epoch 64/200\n",
      "3/3 [==============================] - 1s 186ms/step - loss: 55404.6641 - categorical_accuracy: 0.3412\n",
      "Epoch 65/200\n",
      "3/3 [==============================] - 1s 187ms/step - loss: 122700.0156 - categorical_accuracy: 0.3176\n",
      "Epoch 66/200\n",
      "3/3 [==============================] - 1s 185ms/step - loss: 117221.8828 - categorical_accuracy: 0.3059\n",
      "Epoch 67/200\n",
      "3/3 [==============================] - 1s 197ms/step - loss: 200872.8750 - categorical_accuracy: 0.2941\n",
      "Epoch 68/200\n",
      "3/3 [==============================] - 1s 194ms/step - loss: 887279.6875 - categorical_accuracy: 0.3529\n",
      "Epoch 69/200\n",
      "3/3 [==============================] - 1s 183ms/step - loss: 284885.1562 - categorical_accuracy: 0.3647\n",
      "Epoch 70/200\n",
      "3/3 [==============================] - 1s 190ms/step - loss: 324100.7500 - categorical_accuracy: 0.3647\n",
      "Epoch 71/200\n",
      "3/3 [==============================] - 1s 182ms/step - loss: 163507.3438 - categorical_accuracy: 0.3059\n",
      "Epoch 72/200\n",
      "3/3 [==============================] - 1s 186ms/step - loss: 232204.8438 - categorical_accuracy: 0.2941\n",
      "Epoch 73/200\n",
      "3/3 [==============================] - 1s 193ms/step - loss: 272025.0312 - categorical_accuracy: 0.1765\n",
      "Epoch 74/200\n",
      "3/3 [==============================] - 1s 195ms/step - loss: 264494.5000 - categorical_accuracy: 0.3176\n",
      "Epoch 75/200\n",
      "3/3 [==============================] - 1s 183ms/step - loss: 261781.4375 - categorical_accuracy: 0.3412\n",
      "Epoch 76/200\n",
      "3/3 [==============================] - 1s 184ms/step - loss: 160070.6406 - categorical_accuracy: 0.3647\n",
      "Epoch 77/200\n",
      "3/3 [==============================] - 1s 186ms/step - loss: 63110.0234 - categorical_accuracy: 0.3176\n",
      "Epoch 78/200\n",
      "3/3 [==============================] - 1s 182ms/step - loss: 52113.7812 - categorical_accuracy: 0.2824\n",
      "Epoch 79/200\n",
      "3/3 [==============================] - 1s 187ms/step - loss: 43657.2305 - categorical_accuracy: 0.2941\n",
      "Epoch 80/200\n",
      "3/3 [==============================] - 1s 186ms/step - loss: 22918.3965 - categorical_accuracy: 0.3882\n",
      "Epoch 81/200\n",
      "3/3 [==============================] - 1s 184ms/step - loss: 19972.6035 - categorical_accuracy: 0.3647\n",
      "Epoch 82/200\n",
      "3/3 [==============================] - 1s 204ms/step - loss: 18073.4941 - categorical_accuracy: 0.3529\n",
      "Epoch 83/200\n",
      "3/3 [==============================] - 1s 185ms/step - loss: 44203.9531 - categorical_accuracy: 0.3529\n",
      "Epoch 84/200\n",
      "3/3 [==============================] - 1s 185ms/step - loss: 523418.7812 - categorical_accuracy: 0.3412\n",
      "Epoch 85/200\n",
      "3/3 [==============================] - 1s 184ms/step - loss: 4020871.2500 - categorical_accuracy: 0.2471\n",
      "Epoch 86/200\n",
      "3/3 [==============================] - 1s 189ms/step - loss: 1992426.3750 - categorical_accuracy: 0.3294\n",
      "Epoch 87/200\n",
      "3/3 [==============================] - 1s 187ms/step - loss: 1313570.8750 - categorical_accuracy: 0.3882\n",
      "Epoch 88/200\n",
      "3/3 [==============================] - 1s 188ms/step - loss: 333830.9062 - categorical_accuracy: 0.3176\n",
      "Epoch 89/200\n",
      "3/3 [==============================] - 1s 189ms/step - loss: 1422744.6250 - categorical_accuracy: 0.3529\n",
      "Epoch 90/200\n",
      "3/3 [==============================] - 1s 197ms/step - loss: 871980.0625 - categorical_accuracy: 0.4353\n",
      "Epoch 91/200\n",
      "3/3 [==============================] - 1s 190ms/step - loss: 2231829.0000 - categorical_accuracy: 0.4235\n",
      "Epoch 92/200\n",
      "3/3 [==============================] - 1s 185ms/step - loss: 222695.3438 - categorical_accuracy: 0.4118\n",
      "Epoch 93/200\n",
      "3/3 [==============================] - 1s 184ms/step - loss: 1055162.0000 - categorical_accuracy: 0.2941\n",
      "Epoch 94/200\n",
      "3/3 [==============================] - 1s 186ms/step - loss: 1352582.8750 - categorical_accuracy: 0.4706\n",
      "Epoch 95/200\n",
      "3/3 [==============================] - 1s 183ms/step - loss: 1254774.0000 - categorical_accuracy: 0.2824\n",
      "Epoch 96/200\n",
      "3/3 [==============================] - 1s 186ms/step - loss: 1055732.8750 - categorical_accuracy: 0.3412\n",
      "Epoch 97/200\n",
      "3/3 [==============================] - 1s 205ms/step - loss: 1068897.5000 - categorical_accuracy: 0.3176\n",
      "Epoch 98/200\n",
      "3/3 [==============================] - 1s 183ms/step - loss: 1177082.3750 - categorical_accuracy: 0.3294\n",
      "Epoch 99/200\n",
      "3/3 [==============================] - 1s 192ms/step - loss: 734137.6250 - categorical_accuracy: 0.3176\n",
      "Epoch 100/200\n",
      "3/3 [==============================] - 1s 186ms/step - loss: 856606.1250 - categorical_accuracy: 0.3412\n",
      "Epoch 101/200\n",
      "3/3 [==============================] - 1s 186ms/step - loss: 575793.8125 - categorical_accuracy: 0.4353\n",
      "Epoch 102/200\n",
      "3/3 [==============================] - 1s 210ms/step - loss: 473685.8750 - categorical_accuracy: 0.3176\n",
      "Epoch 103/200\n",
      "3/3 [==============================] - 1s 187ms/step - loss: 245861.9219 - categorical_accuracy: 0.3765\n",
      "Epoch 104/200\n",
      "3/3 [==============================] - 1s 182ms/step - loss: 328766.5000 - categorical_accuracy: 0.2471\n",
      "Epoch 105/200\n",
      "3/3 [==============================] - 1s 182ms/step - loss: 110462.4141 - categorical_accuracy: 0.3176\n",
      "Epoch 106/200\n",
      "3/3 [==============================] - 1s 189ms/step - loss: 50782.2461 - categorical_accuracy: 0.3529\n",
      "Epoch 107/200\n",
      "3/3 [==============================] - 1s 213ms/step - loss: 50113.4062 - categorical_accuracy: 0.3765\n",
      "Epoch 108/200\n",
      "3/3 [==============================] - 1s 184ms/step - loss: 47415.4258 - categorical_accuracy: 0.2706\n",
      "Epoch 109/200\n",
      "3/3 [==============================] - 1s 183ms/step - loss: 32470.6934 - categorical_accuracy: 0.3294\n",
      "Epoch 110/200\n",
      "3/3 [==============================] - 1s 184ms/step - loss: 29774.0117 - categorical_accuracy: 0.3176\n",
      "Epoch 111/200\n",
      "3/3 [==============================] - 1s 181ms/step - loss: 30310.3184 - categorical_accuracy: 0.2588\n",
      "Epoch 112/200\n",
      "3/3 [==============================] - 1s 210ms/step - loss: 55862.6641 - categorical_accuracy: 0.2353\n",
      "Epoch 113/200\n",
      "3/3 [==============================] - 1s 184ms/step - loss: 34457.7930 - categorical_accuracy: 0.3647\n",
      "Epoch 114/200\n",
      "3/3 [==============================] - 1s 183ms/step - loss: 41184.4883 - categorical_accuracy: 0.3176\n",
      "Epoch 115/200\n",
      "3/3 [==============================] - 1s 192ms/step - loss: 40136.6250 - categorical_accuracy: 0.2941\n",
      "Epoch 116/200\n",
      "3/3 [==============================] - 1s 192ms/step - loss: 40517.8867 - categorical_accuracy: 0.2706\n",
      "Epoch 117/200\n",
      "3/3 [==============================] - 1s 184ms/step - loss: 31772.5820 - categorical_accuracy: 0.3529\n",
      "Epoch 118/200\n",
      "3/3 [==============================] - 1s 184ms/step - loss: 22086.7969 - categorical_accuracy: 0.3647\n",
      "Epoch 119/200\n",
      "3/3 [==============================] - 1s 183ms/step - loss: 21465.0918 - categorical_accuracy: 0.4353\n",
      "Epoch 120/200\n",
      "3/3 [==============================] - 1s 210ms/step - loss: 20757.5879 - categorical_accuracy: 0.3294\n",
      "Epoch 121/200\n",
      "3/3 [==============================] - 1s 184ms/step - loss: 22700.1328 - categorical_accuracy: 0.3882\n",
      "Epoch 122/200\n",
      "3/3 [==============================] - 1s 188ms/step - loss: 12472.5439 - categorical_accuracy: 0.3765\n",
      "Epoch 123/200\n",
      "3/3 [==============================] - 1s 207ms/step - loss: 11142.0020 - categorical_accuracy: 0.3412\n",
      "Epoch 124/200\n",
      "3/3 [==============================] - 1s 184ms/step - loss: 16149.8672 - categorical_accuracy: 0.2118\n",
      "Epoch 125/200\n",
      "3/3 [==============================] - 1s 184ms/step - loss: 40700.6172 - categorical_accuracy: 0.3765\n",
      "Epoch 126/200\n",
      "3/3 [==============================] - 1s 195ms/step - loss: 31654.3008 - categorical_accuracy: 0.4000\n",
      "Epoch 127/200\n",
      "3/3 [==============================] - 1s 196ms/step - loss: 17953.7598 - categorical_accuracy: 0.2000\n",
      "Epoch 128/200\n",
      "3/3 [==============================] - 1s 221ms/step - loss: 14556.3672 - categorical_accuracy: 0.2471\n",
      "Epoch 129/200\n",
      "3/3 [==============================] - 1s 189ms/step - loss: 12724.9180 - categorical_accuracy: 0.4471\n",
      "Epoch 130/200\n",
      "3/3 [==============================] - 1s 194ms/step - loss: 14359.5225 - categorical_accuracy: 0.4706\n",
      "Epoch 131/200\n",
      "3/3 [==============================] - 1s 200ms/step - loss: 12288.0342 - categorical_accuracy: 0.3059\n",
      "Epoch 132/200\n",
      "3/3 [==============================] - 1s 222ms/step - loss: 6225.3286 - categorical_accuracy: 0.5294\n",
      "Epoch 133/200\n",
      "3/3 [==============================] - 1s 189ms/step - loss: 7772.4131 - categorical_accuracy: 0.4706\n",
      "Epoch 134/200\n",
      "3/3 [==============================] - 1s 184ms/step - loss: 8494.1777 - categorical_accuracy: 0.5059\n",
      "Epoch 135/200\n",
      "3/3 [==============================] - 1s 187ms/step - loss: 9823.2363 - categorical_accuracy: 0.4235\n",
      "Epoch 136/200\n",
      "3/3 [==============================] - 1s 206ms/step - loss: 8111.9189 - categorical_accuracy: 0.3647\n",
      "Epoch 137/200\n",
      "3/3 [==============================] - 1s 184ms/step - loss: 7405.1914 - categorical_accuracy: 0.4588\n",
      "Epoch 138/200\n",
      "3/3 [==============================] - 1s 185ms/step - loss: 11048.6533 - categorical_accuracy: 0.3412\n",
      "Epoch 139/200\n",
      "3/3 [==============================] - 1s 197ms/step - loss: 33399.4297 - categorical_accuracy: 0.3529\n",
      "Epoch 140/200\n",
      "3/3 [==============================] - 1s 187ms/step - loss: 43818.4336 - categorical_accuracy: 0.3294\n",
      "Epoch 141/200\n",
      "3/3 [==============================] - 1s 218ms/step - loss: 30027.5156 - categorical_accuracy: 0.3529\n",
      "Epoch 142/200\n",
      "3/3 [==============================] - 1s 222ms/step - loss: 17812.9160 - categorical_accuracy: 0.3529\n",
      "Epoch 143/200\n",
      "3/3 [==============================] - 1s 227ms/step - loss: 24721.4316 - categorical_accuracy: 0.3176\n",
      "Epoch 144/200\n",
      "3/3 [==============================] - 1s 217ms/step - loss: 67212.6172 - categorical_accuracy: 0.2706\n",
      "Epoch 145/200\n",
      "3/3 [==============================] - 1s 184ms/step - loss: 57123.9180 - categorical_accuracy: 0.3412\n",
      "Epoch 146/200\n",
      "3/3 [==============================] - 1s 198ms/step - loss: 57308.6016 - categorical_accuracy: 0.3176\n",
      "Epoch 147/200\n",
      "3/3 [==============================] - 1s 208ms/step - loss: 32019.9668 - categorical_accuracy: 0.3882\n",
      "Epoch 148/200\n",
      "3/3 [==============================] - 1s 188ms/step - loss: 34205.6172 - categorical_accuracy: 0.3059\n",
      "Epoch 149/200\n",
      "3/3 [==============================] - 1s 207ms/step - loss: 31080.2227 - categorical_accuracy: 0.3176\n",
      "Epoch 150/200\n",
      "3/3 [==============================] - 1s 187ms/step - loss: 41223.4805 - categorical_accuracy: 0.2941\n",
      "Epoch 151/200\n",
      "3/3 [==============================] - 1s 182ms/step - loss: 34475.1172 - categorical_accuracy: 0.3882\n",
      "Epoch 152/200\n",
      "3/3 [==============================] - 1s 185ms/step - loss: 23054.4746 - categorical_accuracy: 0.4118\n",
      "Epoch 153/200\n",
      "3/3 [==============================] - 1s 214ms/step - loss: 23635.3711 - categorical_accuracy: 0.3882\n",
      "Epoch 154/200\n",
      "3/3 [==============================] - 1s 185ms/step - loss: 22110.9023 - categorical_accuracy: 0.3529\n",
      "Epoch 155/200\n",
      "3/3 [==============================] - 1s 183ms/step - loss: 21422.8711 - categorical_accuracy: 0.4000\n",
      "Epoch 156/200\n",
      "3/3 [==============================] - 1s 192ms/step - loss: 26670.5938 - categorical_accuracy: 0.2471\n",
      "Epoch 157/200\n",
      "3/3 [==============================] - 1s 194ms/step - loss: 19101.3594 - categorical_accuracy: 0.3765\n",
      "Epoch 158/200\n",
      "3/3 [==============================] - 1s 185ms/step - loss: 17326.3906 - categorical_accuracy: 0.4000\n",
      "Epoch 159/200\n",
      "3/3 [==============================] - 1s 211ms/step - loss: 27432.5059 - categorical_accuracy: 0.3412\n",
      "Epoch 160/200\n",
      "3/3 [==============================] - 1s 191ms/step - loss: 19257.8906 - categorical_accuracy: 0.3765\n",
      "Epoch 161/200\n",
      "3/3 [==============================] - 1s 216ms/step - loss: 26384.6387 - categorical_accuracy: 0.3412\n",
      "Epoch 162/200\n",
      "3/3 [==============================] - 1s 187ms/step - loss: 13468.8057 - categorical_accuracy: 0.4235\n",
      "Epoch 163/200\n",
      "3/3 [==============================] - 1s 213ms/step - loss: 20422.6836 - categorical_accuracy: 0.3882\n",
      "Epoch 164/200\n",
      "3/3 [==============================] - 1s 221ms/step - loss: 18394.7617 - categorical_accuracy: 0.4235\n",
      "Epoch 165/200\n",
      "3/3 [==============================] - 1s 211ms/step - loss: 22082.8398 - categorical_accuracy: 0.3294\n",
      "Epoch 166/200\n",
      "3/3 [==============================] - 1s 188ms/step - loss: 18153.2930 - categorical_accuracy: 0.3412\n",
      "Epoch 167/200\n",
      "3/3 [==============================] - 1s 213ms/step - loss: 24949.6387 - categorical_accuracy: 0.2941\n",
      "Epoch 168/200\n",
      "3/3 [==============================] - 1s 186ms/step - loss: 17878.5840 - categorical_accuracy: 0.2471\n",
      "Epoch 169/200\n",
      "3/3 [==============================] - 1s 184ms/step - loss: 16599.4492 - categorical_accuracy: 0.3059\n",
      "Epoch 170/200\n",
      "3/3 [==============================] - 1s 195ms/step - loss: 18134.1777 - categorical_accuracy: 0.3176\n",
      "Epoch 171/200\n",
      "3/3 [==============================] - 1s 187ms/step - loss: 19474.6582 - categorical_accuracy: 0.3176\n",
      "Epoch 172/200\n",
      "3/3 [==============================] - 1s 183ms/step - loss: 21210.4844 - categorical_accuracy: 0.3529\n",
      "Epoch 173/200\n",
      "3/3 [==============================] - 1s 214ms/step - loss: 32694.3125 - categorical_accuracy: 0.2824\n",
      "Epoch 174/200\n",
      "3/3 [==============================] - 1s 185ms/step - loss: 32178.0527 - categorical_accuracy: 0.2235\n",
      "Epoch 175/200\n",
      "3/3 [==============================] - 1s 189ms/step - loss: 23248.5508 - categorical_accuracy: 0.3059\n",
      "Epoch 176/200\n",
      "3/3 [==============================] - 1s 214ms/step - loss: 21329.0879 - categorical_accuracy: 0.4235\n",
      "Epoch 177/200\n",
      "3/3 [==============================] - 1s 187ms/step - loss: 27094.5000 - categorical_accuracy: 0.2588\n",
      "Epoch 178/200\n",
      "3/3 [==============================] - 1s 220ms/step - loss: 18378.9473 - categorical_accuracy: 0.2588\n",
      "Epoch 179/200\n",
      "3/3 [==============================] - 1s 220ms/step - loss: 16631.9180 - categorical_accuracy: 0.4118\n",
      "Epoch 180/200\n",
      "3/3 [==============================] - 1s 222ms/step - loss: 18377.8652 - categorical_accuracy: 0.4824\n",
      "Epoch 181/200\n",
      "3/3 [==============================] - 1s 182ms/step - loss: 26150.7441 - categorical_accuracy: 0.4706\n",
      "Epoch 182/200\n",
      "3/3 [==============================] - 1s 196ms/step - loss: 10624.5176 - categorical_accuracy: 0.2941\n",
      "Epoch 183/200\n",
      "3/3 [==============================] - 1s 188ms/step - loss: 9389.5225 - categorical_accuracy: 0.3412\n",
      "Epoch 184/200\n",
      "3/3 [==============================] - 1s 219ms/step - loss: 7266.8926 - categorical_accuracy: 0.4000\n",
      "Epoch 185/200\n",
      "3/3 [==============================] - 1s 186ms/step - loss: 16855.6367 - categorical_accuracy: 0.3647\n",
      "Epoch 186/200\n",
      "3/3 [==============================] - 1s 218ms/step - loss: 20411.6973 - categorical_accuracy: 0.2706\n",
      "Epoch 187/200\n",
      "3/3 [==============================] - 1s 224ms/step - loss: 19448.0625 - categorical_accuracy: 0.2941\n",
      "Epoch 188/200\n",
      "3/3 [==============================] - 1s 222ms/step - loss: 17263.7715 - categorical_accuracy: 0.2941\n",
      "Epoch 189/200\n",
      "3/3 [==============================] - 1s 188ms/step - loss: 11824.7305 - categorical_accuracy: 0.2000\n",
      "Epoch 190/200\n",
      "3/3 [==============================] - 1s 214ms/step - loss: 11141.7217 - categorical_accuracy: 0.4471\n",
      "Epoch 191/200\n",
      "3/3 [==============================] - 1s 217ms/step - loss: 10377.3848 - categorical_accuracy: 0.2471\n",
      "Epoch 192/200\n",
      "3/3 [==============================] - 1s 220ms/step - loss: 15041.6924 - categorical_accuracy: 0.3765\n",
      "Epoch 193/200\n",
      "3/3 [==============================] - 1s 185ms/step - loss: 23826.0820 - categorical_accuracy: 0.3882\n",
      "Epoch 194/200\n",
      "3/3 [==============================] - 1s 185ms/step - loss: 26408.5820 - categorical_accuracy: 0.3529\n",
      "Epoch 195/200\n",
      "3/3 [==============================] - 1s 183ms/step - loss: 9902.4131 - categorical_accuracy: 0.3882\n",
      "Epoch 196/200\n",
      "3/3 [==============================] - 1s 206ms/step - loss: 10717.2178 - categorical_accuracy: 0.5059\n",
      "Epoch 197/200\n",
      "3/3 [==============================] - 1s 181ms/step - loss: 17718.8379 - categorical_accuracy: 0.3412\n",
      "Epoch 198/200\n",
      "3/3 [==============================] - 1s 188ms/step - loss: 23976.7031 - categorical_accuracy: 0.1294\n",
      "Epoch 199/200\n",
      "3/3 [==============================] - 1s 185ms/step - loss: 10679.2803 - categorical_accuracy: 0.3647\n",
      "Epoch 200/200\n",
      "3/3 [==============================] - 1s 183ms/step - loss: 11567.0400 - categorical_accuracy: 0.4235\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x22329f3ebb0>"
      ]
     },
     "execution_count": 191,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(x_train,y_train,epochs=200)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 192,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_7\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " lstm_21 (LSTM)              (None, 82, 64)            442112    \n",
      "                                                                 \n",
      " lstm_22 (LSTM)              (None, 82, 128)           98816     \n",
      "                                                                 \n",
      " lstm_23 (LSTM)              (None, 64)                49408     \n",
      "                                                                 \n",
      " dense_21 (Dense)            (None, 64)                4160      \n",
      "                                                                 \n",
      " dense_22 (Dense)            (None, 32)                2080      \n",
      "                                                                 \n",
      " dense_23 (Dense)            (None, 3)                 99        \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 596,675\n",
      "Trainable params: 596,675\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 193,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 278ms/step\n"
     ]
    }
   ],
   "source": [
    "res = model.predict(x_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 194,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'aa'"
      ]
     },
     "execution_count": 194,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "actions[np.argmax(res[0])]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 195,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'au'"
      ]
     },
     "execution_count": 195,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "actions[np.argmax(y_test[0])]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 202,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.save('IndHandSigns.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 203,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.load_weights('IndHandSigns.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 204,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import multilabel_confusion_matrix, accuracy_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 205,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 50ms/step\n"
     ]
    }
   ],
   "source": [
    "ypred = model.predict(x_test)\n",
    "ytrue = np.argmax(y_test, axis=1).tolist()\n",
    "ypred = np.argmax(ypred, axis=1).tolist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 206,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[[3, 1],\n",
       "        [1, 0]],\n",
       "\n",
       "       [[1, 2],\n",
       "        [2, 0]],\n",
       "\n",
       "       [[2, 1],\n",
       "        [1, 1]]], dtype=int64)"
      ]
     },
     "execution_count": 206,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "multilabel_confusion_matrix(ytrue, ypred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 207,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.2"
      ]
     },
     "execution_count": 207,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy_score(ytrue, ypred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 208,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 383ms/step\n",
      "1/1 [==============================] - 0s 18ms/step\n",
      "1/1 [==============================] - 0s 18ms/step\n",
      "1/1 [==============================] - 0s 18ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 18ms/step\n",
      "1/1 [==============================] - 0s 20ms/step\n",
      "1/1 [==============================] - 0s 18ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 18ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 27ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 19ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 19ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 18ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 20ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 14ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 18ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 14ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 18ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 18ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 18ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 9ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 19ms/step\n",
      "1/1 [==============================] - 0s 19ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 14ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 10ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 20ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 13ms/step\n",
      "1/1 [==============================] - 0s 14ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 14ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 14ms/step\n",
      "1/1 [==============================] - 0s 14ms/step\n",
      "1/1 [==============================] - 0s 18ms/step\n",
      "1/1 [==============================] - 0s 14ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 14ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 18ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n"
     ]
    }
   ],
   "source": [
    "# 1. New detection variables\n",
    "sequence = []\n",
    "sentence = []\n",
    "predictions = []\n",
    "threshold = 0.7\n",
    "\n",
    "try:\n",
    "    cap = cv2.VideoCapture(0)\n",
    "    # Set mediapipe model\n",
    "    with mp_holistic.Holistic(\n",
    "        min_detection_confidence=0.6, min_tracking_confidence=0.6\n",
    "    ) as holistic:\n",
    "        while cap.isOpened():\n",
    "\n",
    "            # Read camera\n",
    "            ret, frame = cap.read()\n",
    "\n",
    "            # Make detections\n",
    "            image, results = mediapipe_detection(frame, holistic)\n",
    "            # print(results)\n",
    "\n",
    "            # Draw landmarks\n",
    "            draw_styled_landmarks(image, results)\n",
    "\n",
    "            # 2. Prediction logic\n",
    "            keypoints = extract_keypoints(results)\n",
    "            sequence.append(keypoints)\n",
    "            sequence = sequence[-30:]\n",
    "\n",
    "            if len(sequence) == 30:\n",
    "                res = model.predict(np.expand_dims(sequence, axis=0))[0]\n",
    "                # print(actions[np.argmax(res)])\n",
    "                predictions.append(np.argmax(res))\n",
    "\n",
    "                # 3. Vizualization logic\n",
    "                if np.unique(predictions[-1:])[0] == np.argmax(res):\n",
    "                    if res[np.argmax(res)] > threshold:\n",
    "\n",
    "                        word = \"\"\n",
    "                        if len(sentence) > 0:\n",
    "                            if actions[np.argmax(res)] != sentence[-1]:\n",
    "                                sentence.append(actions[np.argmax(res)])\n",
    "                                word = actions[np.argmax(res)]\n",
    "                        else:\n",
    "                            sentence.append(actions[np.argmax(res)])\n",
    "                            word = actions[np.argmax(res)]\n",
    "\n",
    "                        # engine = pyttsx3.init()\n",
    "                        # engine.say(word)\n",
    "                        # # play the speech\n",
    "                        # engine.runAndWait()\n",
    "\n",
    "                if len(sentence) > 1:\n",
    "                    sentence = sentence[-1:]\n",
    "\n",
    "                # Probabilities Vizualization\n",
    "                # image = probability_vizulization(res, actions, image, colors)\n",
    "                \n",
    "                # cv2.putText(image, 'PROB', (15,12), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (0, 0, 0), 1, cv2.LINE_AA)\n",
    "                # cv2.putText(image, str(round(body_language_prob[np.argmax(body_language_prob)],2)) , (10,40), cv2.FONT_HERSHEY_SIMPLEX, 1, (255, 255, 255), 2, cv2.LINE_AA)\n",
    "\n",
    "            cv2.rectangle(image, (0, 0), (640, 40), (245, 117, 16), -1)\n",
    "            cv2.putText(\n",
    "                image,\n",
    "                \" \".join(sentence),\n",
    "                (3, 30),\n",
    "                cv2.FONT_HERSHEY_SIMPLEX,\n",
    "                1,\n",
    "                (255, 255, 255),\n",
    "                2,\n",
    "                cv2.LINE_AA,\n",
    "            )\n",
    "            \n",
    "            # Flip the image\n",
    "            # image = cv2.flip(image, 1)\n",
    "\n",
    "            # Show to screen\n",
    "            cv2.imshow(\"OpenCV Feed\", image)\n",
    "\n",
    "            # Break gracefully\n",
    "            if cv2.waitKey(10) & 0xFF == ord(\"q\"):\n",
    "                break\n",
    "        cap.release()\n",
    "        cv2.destroyAllWindows()\n",
    "\n",
    "finally:\n",
    "    cap.release()\n",
    "    cv2.destroyAllWindows()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}