building_xg_boost_model.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import xgboost as xgb\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import accuracy_score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from utils.image_to_embedding import get_image_to_embedding\n",
    "from utils.static_check import get_static_difference\n",
    "from utils.optical_flow_check import get_optical_flow\n",
    "from diffusers import AutoencoderKL\n",
    "import torch\n",
    "\n",
    "import cv2\n",
    "import numpy as np\n",
    "from PIL import Image\n",
    "import os\n",
    "from torchvision import transforms\n",
    "\n",
    "def image_transform(image):\n",
    "    transform = transforms.Compose([\n",
    "        transforms.Resize((320, 240)),\n",
    "        transforms.CenterCrop(240),\n",
    "        transforms.ToTensor(),  \n",
    "    ])\n",
    "    return transform(image).unsqueeze(0)\n",
    "\n",
    "def load_image(folder_path):\n",
    "    # Sort the frames\n",
    "    frames = sorted(os.listdir(folder_path))\n",
    "    conv_frames = []\n",
    "    \n",
    "    # Convert the frames to tensor\n",
    "    for frame in frames:\n",
    "        image = Image.open(os.path.join(folder_path, frame)).convert('RGB')\n",
    "        image = image_transform(image)\n",
    "        conv_frames.append(image)\n",
    "    return conv_frames\n",
    "\n",
    "def get_model():\n",
    "    model = AutoencoderKL.from_pretrained(\"stabilityai/sdxl-vae\")\n",
    "    return model\n",
    "\n",
    "def get_inference(folder_path, model, device):\n",
    "    frames = load_image(folder_path)\n",
    "\n",
    "    # Getting static difference\n",
    "    rgb_diff_max, rgb_diff_mean = get_static_difference(frames)\n",
    "\n",
    "    # getting optical flow\n",
    "    max_velocity, avg_velocity = get_optical_flow(frames)\n",
    "    \n",
    "    # Getting image context similarity\n",
    "    frame_mse, frame_cos_sim = get_image_to_embedding(frames, model, device)\n",
    "\n",
    "    # Getting number of frames\n",
    "    num_frames = len(frames)\n",
    "\n",
    "    return rgb_diff_max, rgb_diff_mean, max_velocity, avg_velocity, frame_mse, frame_cos_sim, num_frames"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load from annotations.json\n",
    "import json\n",
    "with open('annotations.json', 'r') as f:\n",
    "    labeled_data = json.load(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "rgb_diff_maxs = []\n",
    "rgb_diff_means = []\n",
    "max_velocities = []\n",
    "avg_velocities = []\n",
    "frame_mses = []\n",
    "frame_cos_sims = []\n",
    "num_frames_list = []\n",
    "labels = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Enumerated 1 paths\n",
      "Enumerated 2 paths\n",
      "Enumerated 3 paths\n",
      "Enumerated 4 paths\n",
      "Enumerated 5 paths\n",
      "Enumerated 6 paths\n",
      "Enumerated 7 paths\n",
      "Enumerated 8 paths\n",
      "Enumerated 9 paths\n",
      "Enumerated 10 paths\n",
      "Enumerated 11 paths\n",
      "Enumerated 12 paths\n",
      "Enumerated 13 paths\n",
      "Enumerated 14 paths\n",
      "Enumerated 15 paths\n",
      "Enumerated 16 paths\n",
      "Enumerated 17 paths\n",
      "Enumerated 18 paths\n",
      "Enumerated 19 paths\n",
      "Enumerated 20 paths\n",
      "Enumerated 21 paths\n",
      "Enumerated 22 paths\n",
      "Enumerated 23 paths\n",
      "Enumerated 24 paths\n",
      "Enumerated 25 paths\n",
      "Enumerated 26 paths\n",
      "Enumerated 27 paths\n",
      "Enumerated 28 paths\n",
      "Enumerated 29 paths\n",
      "Enumerated 30 paths\n",
      "Enumerated 31 paths\n",
      "Enumerated 32 paths\n",
      "Enumerated 33 paths\n",
      "Enumerated 34 paths\n",
      "Enumerated 35 paths\n",
      "Enumerated 36 paths\n",
      "Enumerated 37 paths\n",
      "Enumerated 38 paths\n",
      "Enumerated 39 paths\n",
      "Enumerated 40 paths\n",
      "Enumerated 41 paths\n",
      "Enumerated 42 paths\n",
      "Enumerated 43 paths\n",
      "Enumerated 44 paths\n",
      "Enumerated 45 paths\n",
      "Enumerated 46 paths\n",
      "Enumerated 47 paths\n",
      "Enumerated 48 paths\n",
      "Enumerated 49 paths\n",
      "Enumerated 50 paths\n",
      "Enumerated 51 paths\n",
      "Enumerated 52 paths\n",
      "Enumerated 53 paths\n",
      "Enumerated 54 paths\n",
      "Enumerated 55 paths\n",
      "Enumerated 56 paths\n",
      "Enumerated 57 paths\n",
      "Enumerated 58 paths\n",
      "Enumerated 59 paths\n",
      "Enumerated 60 paths\n",
      "Enumerated 61 paths\n",
      "Enumerated 62 paths\n",
      "Enumerated 63 paths\n",
      "Enumerated 64 paths\n",
      "Enumerated 65 paths\n",
      "Enumerated 66 paths\n",
      "Enumerated 67 paths\n",
      "Enumerated 68 paths\n",
      "Enumerated 69 paths\n",
      "Enumerated 70 paths\n",
      "Enumerated 71 paths\n",
      "Enumerated 72 paths\n",
      "Enumerated 73 paths\n",
      "Enumerated 74 paths\n",
      "Enumerated 75 paths\n",
      "Enumerated 76 paths\n",
      "Enumerated 77 paths\n",
      "Enumerated 78 paths\n",
      "Enumerated 79 paths\n",
      "Enumerated 80 paths\n",
      "Enumerated 81 paths\n",
      "Enumerated 82 paths\n",
      "Enumerated 83 paths\n",
      "Enumerated 84 paths\n",
      "Enumerated 85 paths\n",
      "Enumerated 86 paths\n",
      "Enumerated 87 paths\n",
      "Enumerated 88 paths\n",
      "Enumerated 89 paths\n",
      "Enumerated 90 paths\n",
      "Enumerated 91 paths\n",
      "Enumerated 92 paths\n",
      "Enumerated 93 paths\n",
      "Enumerated 94 paths\n",
      "Enumerated 95 paths\n",
      "Enumerated 96 paths\n",
      "Enumerated 97 paths\n",
      "Enumerated 98 paths\n",
      "Enumerated 99 paths\n",
      "Enumerated 100 paths\n",
      "Enumerated 101 paths\n",
      "Enumerated 102 paths\n",
      "Enumerated 103 paths\n",
      "Enumerated 104 paths\n",
      "Enumerated 105 paths\n",
      "Enumerated 106 paths\n",
      "Enumerated 107 paths\n",
      "Enumerated 108 paths\n",
      "Enumerated 109 paths\n",
      "Enumerated 110 paths\n",
      "Enumerated 111 paths\n",
      "Enumerated 112 paths\n",
      "Enumerated 113 paths\n",
      "Enumerated 114 paths\n",
      "Enumerated 115 paths\n",
      "Enumerated 116 paths\n",
      "Enumerated 117 paths\n",
      "Enumerated 118 paths\n",
      "Enumerated 119 paths\n",
      "Enumerated 120 paths\n",
      "Enumerated 121 paths\n",
      "Enumerated 122 paths\n",
      "Enumerated 123 paths\n",
      "Enumerated 124 paths\n",
      "Enumerated 125 paths\n",
      "Enumerated 126 paths\n",
      "Enumerated 127 paths\n",
      "Enumerated 128 paths\n",
      "Enumerated 129 paths\n",
      "Enumerated 130 paths\n",
      "Enumerated 131 paths\n",
      "Enumerated 132 paths\n",
      "Enumerated 133 paths\n",
      "Enumerated 134 paths\n",
      "Enumerated 135 paths\n",
      "Enumerated 136 paths\n",
      "Enumerated 137 paths\n",
      "Enumerated 138 paths\n",
      "Enumerated 139 paths\n",
      "Enumerated 140 paths\n",
      "Enumerated 141 paths\n",
      "Enumerated 142 paths\n",
      "Enumerated 143 paths\n",
      "Enumerated 144 paths\n",
      "Enumerated 145 paths\n",
      "Enumerated 146 paths\n",
      "Enumerated 147 paths\n",
      "Enumerated 148 paths\n",
      "Enumerated 149 paths\n",
      "Enumerated 150 paths\n",
      "Enumerated 151 paths\n",
      "Enumerated 152 paths\n",
      "Enumerated 153 paths\n",
      "Enumerated 154 paths\n",
      "Enumerated 155 paths\n",
      "Enumerated 156 paths\n",
      "Enumerated 157 paths\n",
      "Enumerated 158 paths\n",
      "Enumerated 159 paths\n",
      "Enumerated 160 paths\n",
      "Enumerated 161 paths\n",
      "Enumerated 162 paths\n",
      "Enumerated 163 paths\n",
      "Enumerated 164 paths\n",
      "Enumerated 165 paths\n",
      "Enumerated 166 paths\n",
      "Enumerated 167 paths\n",
      "Enumerated 168 paths\n",
      "Enumerated 169 paths\n",
      "Enumerated 170 paths\n",
      "Enumerated 171 paths\n",
      "Enumerated 172 paths\n",
      "Enumerated 173 paths\n",
      "Enumerated 174 paths\n",
      "Enumerated 175 paths\n",
      "Enumerated 176 paths\n",
      "Enumerated 177 paths\n",
      "Enumerated 178 paths\n",
      "Enumerated 179 paths\n",
      "Enumerated 180 paths\n",
      "Enumerated 181 paths\n",
      "Enumerated 182 paths\n",
      "Enumerated 183 paths\n",
      "Enumerated 184 paths\n",
      "Enumerated 185 paths\n",
      "Enumerated 186 paths\n",
      "Enumerated 187 paths\n",
      "Enumerated 188 paths\n",
      "Enumerated 189 paths\n",
      "Enumerated 190 paths\n",
      "Enumerated 191 paths\n",
      "Enumerated 192 paths\n",
      "Enumerated 193 paths\n",
      "Enumerated 194 paths\n",
      "Enumerated 195 paths\n",
      "Enumerated 196 paths\n",
      "Enumerated 197 paths\n",
      "Enumerated 198 paths\n",
      "Enumerated 199 paths\n",
      "Enumerated 200 paths\n",
      "Enumerated 201 paths\n",
      "Enumerated 202 paths\n",
      "Enumerated 203 paths\n",
      "Enumerated 204 paths\n",
      "Enumerated 205 paths\n",
      "Enumerated 206 paths\n",
      "Enumerated 207 paths\n",
      "Enumerated 208 paths\n",
      "Enumerated 209 paths\n",
      "Enumerated 210 paths\n",
      "Enumerated 211 paths\n",
      "Enumerated 212 paths\n",
      "Enumerated 213 paths\n",
      "Enumerated 214 paths\n",
      "Enumerated 215 paths\n",
      "Enumerated 216 paths\n",
      "Enumerated 217 paths\n",
      "Enumerated 218 paths\n",
      "Enumerated 219 paths\n",
      "Enumerated 220 paths\n",
      "Enumerated 221 paths\n",
      "Enumerated 222 paths\n",
      "Enumerated 223 paths\n",
      "Enumerated 224 paths\n",
      "Enumerated 225 paths\n",
      "Enumerated 226 paths\n",
      "Enumerated 227 paths\n",
      "Enumerated 228 paths\n",
      "Enumerated 229 paths\n",
      "Enumerated 230 paths\n",
      "Enumerated 231 paths\n",
      "Enumerated 232 paths\n",
      "Enumerated 233 paths\n",
      "Enumerated 234 paths\n",
      "Enumerated 235 paths\n",
      "Enumerated 236 paths\n",
      "Enumerated 237 paths\n",
      "Enumerated 238 paths\n",
      "Enumerated 239 paths\n",
      "Enumerated 240 paths\n",
      "Enumerated 241 paths\n",
      "Enumerated 242 paths\n",
      "Enumerated 243 paths\n",
      "Enumerated 244 paths\n",
      "Enumerated 245 paths\n",
      "Enumerated 246 paths\n",
      "Enumerated 247 paths\n",
      "Enumerated 248 paths\n",
      "Enumerated 249 paths\n",
      "Enumerated 250 paths\n",
      "Enumerated 251 paths\n",
      "Enumerated 252 paths\n",
      "Enumerated 253 paths\n",
      "Enumerated 254 paths\n",
      "Enumerated 255 paths\n",
      "Enumerated 256 paths\n",
      "Enumerated 257 paths\n",
      "Enumerated 258 paths\n"
     ]
    }
   ],
   "source": [
    "device = \"cuda\" if torch.cuda.is_available() else \"cpu\"\n",
    "model = get_model().to(device)\n",
    "\n",
    "for i, path in enumerate(labeled_data.keys()):\n",
    "    rgb_diff_max, rgb_diff_mean, max_velocity, avg_velocity, frame_mse, frame_cos_sim, num_frames = get_inference(path, model, device)\n",
    "    rgb_diff_maxs.append(rgb_diff_max)\n",
    "    rgb_diff_means.append(rgb_diff_mean)\n",
    "    max_velocities.append(max_velocity)\n",
    "    avg_velocities.append(avg_velocity)\n",
    "    frame_mses.append(frame_mse)\n",
    "    frame_cos_sims.append(frame_cos_sim)\n",
    "    num_frames_list.append(num_frames)\n",
    "    labels.append(labeled_data[path])\n",
    "    print(f'Enumerated {i+1} paths')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "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>max_rgb_diff</th>\n",
       "      <th>mean_rgb_diff</th>\n",
       "      <th>max_velocity</th>\n",
       "      <th>mean_velocity</th>\n",
       "      <th>mse</th>\n",
       "      <th>cos_sim</th>\n",
       "      <th>no_frames</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>696.234253</td>\n",
       "      <td>265.932800</td>\n",
       "      <td>12.473114</td>\n",
       "      <td>7.542836</td>\n",
       "      <td>11.076690</td>\n",
       "      <td>0.748877</td>\n",
       "      <td>60</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>142.426315</td>\n",
       "      <td>103.156654</td>\n",
       "      <td>0.620393</td>\n",
       "      <td>0.525993</td>\n",
       "      <td>3.880109</td>\n",
       "      <td>0.924669</td>\n",
       "      <td>7</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1034.846191</td>\n",
       "      <td>390.833527</td>\n",
       "      <td>5.756709</td>\n",
       "      <td>4.963755</td>\n",
       "      <td>13.054274</td>\n",
       "      <td>0.721435</td>\n",
       "      <td>59</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>104.106194</td>\n",
       "      <td>20.794767</td>\n",
       "      <td>0.917800</td>\n",
       "      <td>0.596388</td>\n",
       "      <td>2.001047</td>\n",
       "      <td>0.976434</td>\n",
       "      <td>26</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5661.599121</td>\n",
       "      <td>367.071655</td>\n",
       "      <td>9.760819</td>\n",
       "      <td>5.455320</td>\n",
       "      <td>11.926335</td>\n",
       "      <td>0.748189</td>\n",
       "      <td>28</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   max_rgb_diff  mean_rgb_diff  max_velocity  mean_velocity        mse  \\\n",
       "0    696.234253     265.932800     12.473114       7.542836  11.076690   \n",
       "1    142.426315     103.156654      0.620393       0.525993   3.880109   \n",
       "2   1034.846191     390.833527      5.756709       4.963755  13.054274   \n",
       "3    104.106194      20.794767      0.917800       0.596388   2.001047   \n",
       "4   5661.599121     367.071655      9.760819       5.455320  11.926335   \n",
       "\n",
       "    cos_sim  no_frames  label  \n",
       "0  0.748877         60      1  \n",
       "1  0.924669          7      1  \n",
       "2  0.721435         59      1  \n",
       "3  0.976434         26      1  \n",
       "4  0.748189         28      0  "
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "\n",
    "dataset = pd.DataFrame({\n",
    "    \"max_rgb_diff\": rgb_diff_maxs,\n",
    "    \"mean_rgb_diff\": rgb_diff_means,\n",
    "    \"max_velocity\": max_velocities,\n",
    "    \"mean_velocity\": avg_velocities,\n",
    "    \"mse\": frame_mses,\n",
    "    \"cos_sim\": frame_cos_sims,\n",
    "    \"no_frames\": num_frames_list,\n",
    "    \"label\": labels\n",
    "})\n",
    "dataset.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "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>max_rgb_diff</th>\n",
       "      <th>mean_rgb_diff</th>\n",
       "      <th>max_velocity</th>\n",
       "      <th>mean_velocity</th>\n",
       "      <th>mse</th>\n",
       "      <th>cos_sim</th>\n",
       "      <th>no_frames</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>max_rgb_diff</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.558016</td>\n",
       "      <td>0.296977</td>\n",
       "      <td>0.158794</td>\n",
       "      <td>0.276107</td>\n",
       "      <td>-0.173854</td>\n",
       "      <td>0.092808</td>\n",
       "      <td>-0.288155</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean_rgb_diff</th>\n",
       "      <td>0.558016</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.324095</td>\n",
       "      <td>0.340854</td>\n",
       "      <td>0.822956</td>\n",
       "      <td>-0.611652</td>\n",
       "      <td>-0.064194</td>\n",
       "      <td>0.023618</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max_velocity</th>\n",
       "      <td>0.296977</td>\n",
       "      <td>0.324095</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.890355</td>\n",
       "      <td>0.401765</td>\n",
       "      <td>-0.358556</td>\n",
       "      <td>0.201489</td>\n",
       "      <td>-0.038308</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean_velocity</th>\n",
       "      <td>0.158794</td>\n",
       "      <td>0.340854</td>\n",
       "      <td>0.890355</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.477395</td>\n",
       "      <td>-0.432766</td>\n",
       "      <td>0.138010</td>\n",
       "      <td>0.048072</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mse</th>\n",
       "      <td>0.276107</td>\n",
       "      <td>0.822956</td>\n",
       "      <td>0.401765</td>\n",
       "      <td>0.477395</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>-0.874607</td>\n",
       "      <td>-0.052466</td>\n",
       "      <td>0.184298</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>cos_sim</th>\n",
       "      <td>-0.173854</td>\n",
       "      <td>-0.611652</td>\n",
       "      <td>-0.358556</td>\n",
       "      <td>-0.432766</td>\n",
       "      <td>-0.874607</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.045450</td>\n",
       "      <td>-0.149141</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>no_frames</th>\n",
       "      <td>0.092808</td>\n",
       "      <td>-0.064194</td>\n",
       "      <td>0.201489</td>\n",
       "      <td>0.138010</td>\n",
       "      <td>-0.052466</td>\n",
       "      <td>0.045450</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.084073</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>label</th>\n",
       "      <td>-0.288155</td>\n",
       "      <td>0.023618</td>\n",
       "      <td>-0.038308</td>\n",
       "      <td>0.048072</td>\n",
       "      <td>0.184298</td>\n",
       "      <td>-0.149141</td>\n",
       "      <td>0.084073</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "               max_rgb_diff  mean_rgb_diff  max_velocity  mean_velocity  \\\n",
       "max_rgb_diff       1.000000       0.558016      0.296977       0.158794   \n",
       "mean_rgb_diff      0.558016       1.000000      0.324095       0.340854   \n",
       "max_velocity       0.296977       0.324095      1.000000       0.890355   \n",
       "mean_velocity      0.158794       0.340854      0.890355       1.000000   \n",
       "mse                0.276107       0.822956      0.401765       0.477395   \n",
       "cos_sim           -0.173854      -0.611652     -0.358556      -0.432766   \n",
       "no_frames          0.092808      -0.064194      0.201489       0.138010   \n",
       "label             -0.288155       0.023618     -0.038308       0.048072   \n",
       "\n",
       "                    mse   cos_sim  no_frames     label  \n",
       "max_rgb_diff   0.276107 -0.173854   0.092808 -0.288155  \n",
       "mean_rgb_diff  0.822956 -0.611652  -0.064194  0.023618  \n",
       "max_velocity   0.401765 -0.358556   0.201489 -0.038308  \n",
       "mean_velocity  0.477395 -0.432766   0.138010  0.048072  \n",
       "mse            1.000000 -0.874607  -0.052466  0.184298  \n",
       "cos_sim       -0.874607  1.000000   0.045450 -0.149141  \n",
       "no_frames     -0.052466  0.045450   1.000000  0.084073  \n",
       "label          0.184298 -0.149141   0.084073  1.000000  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dataset.corr()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "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>max_rgb_diff</th>\n",
       "      <th>mean_rgb_diff</th>\n",
       "      <th>max_velocity</th>\n",
       "      <th>mean_velocity</th>\n",
       "      <th>mse</th>\n",
       "      <th>cos_sim</th>\n",
       "      <th>no_frames</th>\n",
       "      <th>label</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>258.000000</td>\n",
       "      <td>258.000000</td>\n",
       "      <td>258.000000</td>\n",
       "      <td>258.000000</td>\n",
       "      <td>258.000000</td>\n",
       "      <td>258.000000</td>\n",
       "      <td>258.000000</td>\n",
       "      <td>258.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>3181.503418</td>\n",
       "      <td>522.466919</td>\n",
       "      <td>7.323443</td>\n",
       "      <td>4.223952</td>\n",
       "      <td>11.293567</td>\n",
       "      <td>0.854618</td>\n",
       "      <td>37.562016</td>\n",
       "      <td>0.616279</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>4735.148438</td>\n",
       "      <td>701.674683</td>\n",
       "      <td>5.824468</td>\n",
       "      <td>3.921226</td>\n",
       "      <td>9.689132</td>\n",
       "      <td>0.132447</td>\n",
       "      <td>29.032218</td>\n",
       "      <td>0.487236</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.109219</td>\n",
       "      <td>0.019441</td>\n",
       "      <td>0.008891</td>\n",
       "      <td>0.007581</td>\n",
       "      <td>0.055443</td>\n",
       "      <td>0.093587</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>342.513474</td>\n",
       "      <td>87.631962</td>\n",
       "      <td>3.136222</td>\n",
       "      <td>1.423700</td>\n",
       "      <td>3.899280</td>\n",
       "      <td>0.792531</td>\n",
       "      <td>20.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>1453.235596</td>\n",
       "      <td>271.241714</td>\n",
       "      <td>6.422338</td>\n",
       "      <td>3.434513</td>\n",
       "      <td>8.828903</td>\n",
       "      <td>0.897889</td>\n",
       "      <td>31.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>4021.528015</td>\n",
       "      <td>653.851608</td>\n",
       "      <td>10.090562</td>\n",
       "      <td>6.312151</td>\n",
       "      <td>15.612548</td>\n",
       "      <td>0.953994</td>\n",
       "      <td>47.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>34677.078125</td>\n",
       "      <td>4575.578125</td>\n",
       "      <td>44.851021</td>\n",
       "      <td>30.878588</td>\n",
       "      <td>49.426112</td>\n",
       "      <td>0.999651</td>\n",
       "      <td>256.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       max_rgb_diff  mean_rgb_diff  max_velocity  mean_velocity         mse  \\\n",
       "count    258.000000     258.000000    258.000000     258.000000  258.000000   \n",
       "mean    3181.503418     522.466919      7.323443       4.223952   11.293567   \n",
       "std     4735.148438     701.674683      5.824468       3.921226    9.689132   \n",
       "min        0.109219       0.019441      0.008891       0.007581    0.055443   \n",
       "25%      342.513474      87.631962      3.136222       1.423700    3.899280   \n",
       "50%     1453.235596     271.241714      6.422338       3.434513    8.828903   \n",
       "75%     4021.528015     653.851608     10.090562       6.312151   15.612548   \n",
       "max    34677.078125    4575.578125     44.851021      30.878588   49.426112   \n",
       "\n",
       "          cos_sim   no_frames       label  \n",
       "count  258.000000  258.000000  258.000000  \n",
       "mean     0.854618   37.562016    0.616279  \n",
       "std      0.132447   29.032218    0.487236  \n",
       "min      0.093587    4.000000    0.000000  \n",
       "25%      0.792531   20.000000    0.000000  \n",
       "50%      0.897889   31.000000    1.000000  \n",
       "75%      0.953994   47.000000    1.000000  \n",
       "max      0.999651  256.000000    1.000000  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dataset.describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "X = dataset.drop(columns=[\"label\"])\n",
    "y = dataset[\"label\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Accuracy: 0.9854368932038835\n",
      "Accuracy: 0.8461538461538461\n"
     ]
    }
   ],
   "source": [
    "# Splitting the data into training and testing sets\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=0)\n",
    "\n",
    "# Creating an XGBoost classifier\n",
    "model = xgb.XGBClassifier()\n",
    "\n",
    "# Training the model\n",
    "eval_set = [(X_test, y_test)]\n",
    "model.fit(X_train, y_train, early_stopping_rounds=10, eval_metric=\"logloss\", eval_set=eval_set, verbose=False)\n",
    "\n",
    "y_pred_train = model.predict(X_train)\n",
    "accuracy_train = accuracy_score(y_train, y_pred_train)\n",
    "print(\"Train Accuracy:\", accuracy_train)\n",
    "\n",
    "# Making predictions on the test set\n",
    "y_pred = model.predict(X_test)\n",
    "\n",
    "# Evaluating the model\n",
    "accuracy = accuracy_score(y_test, y_pred)\n",
    "print(\"Accuracy:\", accuracy)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pickle\n",
    "\n",
    "# Save the trained model to a file\n",
    "filename = 'xgboost_model.pkl'\n",
    "pickle.dump(model, open(filename, 'wb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy of loaded model: 0.8461538461538461\n"
     ]
    }
   ],
   "source": [
    "# TO TEST LOADING THE SAVED MODEL\n",
    "\n",
    "# Load the saved model from file\n",
    "classifier_model = pickle.load(open(filename, 'rb'))\n",
    "\n",
    "# Use the loaded model for predictions\n",
    "y_pred_loaded = classifier_model.predict(X_test)\n",
    "\n",
    "# Evaluate the loaded model\n",
    "accuracy_loaded = accuracy_score(y_test, y_pred_loaded)\n",
    "print(\"Accuracy of loaded model:\", accuracy_loaded)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "mlda",
   "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": "3.9.18"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}