info_classifier_model.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using cache found in C:\\Users\\virgi/.cache\\torch\\hub\\ultralytics_yolov5_master\n",
      "\u001b[31m\u001b[1mrequirements:\u001b[0m pandas>=1.1.4 not found and is required by YOLOv5, attempting auto-update...\n",
      "\u001b[31m\u001b[1mrequirements:\u001b[0m Command 'pip install 'pandas>=1.1.4' ' returned non-zero exit status 1.\n",
      "YOLOv5  2022-5-16 Python-3.7.8 torch-1.8.1+cu102 CUDA:0 (NVIDIA GeForce RTX 2060, 6144MiB)\n",
      "\n",
      "Fusing layers... \n",
      "YOLOv5s summary: 213 layers, 7225885 parameters, 0 gradients\n",
      "Adding AutoShape... \n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import pandas as pd\n",
    "import cv2\n",
    "from detector import get_info\n",
    "import matplotlib.pyplot as plt\n",
    "import tensorflow as tf\n",
    "import numpy as np\n",
    "from sklearn.metrics import confusion_matrix\n",
    "import seaborn as sns\n",
    "import matplotlib\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_folder = 'results/models/model_big_with_hand'\n",
    "model = tf.keras.models.load_model(model_folder)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>user</th>\n",
       "      <th>sex</th>\n",
       "      <th>sign</th>\n",
       "      <th>frame</th>\n",
       "      <th>hand</th>\n",
       "      <th>...</th>\n",
       "      <th>hand_mark_19z</th>\n",
       "      <th>hand_mark_20x</th>\n",
       "      <th>hand_mark_20y</th>\n",
       "      <th>hand_mark_20z</th>\n",
       "      <th>hand_id</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>user01</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>4</td>\n",
       "      <td>Right</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.084381</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.348559</td>\n",
       "      <td>-0.089569</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>user01</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>2</td>\n",
       "      <td>Right</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.070153</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.274275</td>\n",
       "      <td>-0.073769</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>user01</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>24</td>\n",
       "      <td>Right</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.050475</td>\n",
       "      <td>0.470683</td>\n",
       "      <td>0.976509</td>\n",
       "      <td>-0.042765</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>user01</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "      <td>Right</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.079948</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.341380</td>\n",
       "      <td>-0.085815</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>user01</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>23</td>\n",
       "      <td>Right</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.052067</td>\n",
       "      <td>0.495761</td>\n",
       "      <td>0.972848</td>\n",
       "      <td>-0.044382</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 70 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "     user  sex  sign  frame   hand  ... hand_mark_19z  hand_mark_20x  \\\n",
       "0  user01    0     5      4  Right  ...     -0.084381       1.000000   \n",
       "1  user01    0     5      2  Right  ...     -0.070153       1.000000   \n",
       "2  user01    0     1     24  Right  ...     -0.050475       0.470683   \n",
       "3  user01    0     5      3  Right  ...     -0.079948       1.000000   \n",
       "4  user01    0     1     23  Right  ...     -0.052067       0.495761   \n",
       "\n",
       "   hand_mark_20y  hand_mark_20z  hand_id  \n",
       "0       0.348559      -0.089569        0  \n",
       "1       0.274275      -0.073769        0  \n",
       "2       0.976509      -0.042765        0  \n",
       "3       0.341380      -0.085815        0  \n",
       "4       0.972848      -0.044382        0  \n",
       "\n",
       "[5 rows x 70 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "marks_names = ['hand_mark_'+str(i).zfill(2)+axis for i in range(21) for axis in ['x','y','z']]\n",
    "train_labels = marks_names + ['hand_id']\n",
    "\n",
    "users = pd.read_csv('users.csv')\n",
    "train_ids = users[users['train']==1]['user_id'].values\n",
    "\n",
    "f_name= 'results/hand_landmarks.csv'\n",
    "data = pd.read_csv(f_name)\n",
    "data['hand_id'] = np.where(data['hand']=='Left', 1, 0)\n",
    "data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_data(labels):  \n",
    "\n",
    "    train_data = data[data['user'].isin(train_ids)]\n",
    "    test_data = data[~data['user'].isin(train_ids)]\n",
    "\n",
    "    X_train = train_data[labels].values\n",
    "    X_test = test_data[labels].values\n",
    "\n",
    "    y_train = train_data['sign'].values -1\n",
    "    y_test = test_data['sign'].values -1\n",
    "    \n",
    "    return X_train, X_test, y_train, y_test\n",
    "\n",
    "def get_model(input_size, output_size):\n",
    "    model = keras.Sequential(\n",
    "        [\n",
    "            keras.Input(shape=(input_size,)),\n",
    "            #layers.Dropout(0.1),\n",
    "            layers.Dense(21, activation=\"relu\"),\n",
    "            #layers.Dropout(0.1),\n",
    "            layers.Dense(10, activation=\"relu\"),\n",
    "            layers.Dense(output_size, activation='softmax'),\n",
    "        ]\n",
    "    )\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train, X_test, y_train, y_test = get_data(marks_names + ['hand_id'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "predictions = model.predict(X_train, verbose=False)\n",
    "y_pred_train = predictions.argmax(axis=-1)\n",
    "\n",
    "predictions = model.predict(X_test, verbose=False)\n",
    "y_pred_test = predictions.argmax(axis=-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_matrix = confusion_matrix(y_train, y_pred_train)\n",
    "test_matrix = confusion_matrix(y_test, y_pred_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x432 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "labels = ['Sign '+str(i) for i in range(1,8)]\n",
    "\n",
    "fig, ax = plt.subplots(1,2,figsize=(14, 6))\n",
    "\n",
    "sns.heatmap(train_matrix, annot=True, cmap=\"YlOrBr\", square=True, ax=ax[0], xticklabels=labels, yticklabels=labels, fmt='g')\n",
    "ax[0].set_yticklabels(labels, rotation=0)\n",
    "ax[0].set_xticklabels(labels, rotation=45)\n",
    "ax[0].set_ylabel('Real category')\n",
    "ax[0].set_xlabel('Predicted category')\n",
    "ax[0].set_title('Train data')\n",
    "sns.heatmap(test_matrix, annot=True, cmap=\"YlOrBr\", square=True, ax=ax[1], xticklabels=labels, yticklabels=labels, fmt='g')\n",
    "ax[1].set_yticklabels(labels, rotation=0)\n",
    "ax[1].set_xticklabels(labels, rotation=45)\n",
    "ax[1].set_ylabel('Real category')\n",
    "ax[1].set_xlabel('Predicted category')\n",
    "ax[1].set_title('Test data')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# train with class_weight"
   ]
  }
 ],
 "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": "3.7.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}