04_whale_detection.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Adapted from radekosmulski [notebook](https://github.com/radekosmulski/whale/blob/master/first_submission.ipynb) for learning purpose on the playground competition instead of the original."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[K     |████████████████████████████████| 727kB 12.2MB/s \n",
      "\u001b[K     |████████████████████████████████| 1.1MB 59.2MB/s \n",
      "\u001b[K     |████████████████████████████████| 51kB 9.2MB/s \n",
      "\u001b[K     |████████████████████████████████| 194kB 66.8MB/s \n",
      "\u001b[K     |████████████████████████████████| 61kB 10.3MB/s \n",
      "\u001b[?25hMounted at /content/gdrive\n"
     ]
    }
   ],
   "source": [
    "!pip install -Uqq fastbook\n",
    "import fastbook\n",
    "fastbook.setup_book()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "root='/content/gdrive/MyDrive/'\n",
    "base=f'{root}/Colab Notebooks/whale'"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Setup Kaggle and Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "kaggle.json\n"
     ]
    }
   ],
   "source": [
    "#Setup kaggle key\n",
    "!mkdir -p ~/.kaggle\n",
    "!cp '{base}/kaggle.json' ~/.kaggle/\n",
    "!ls ~/.kaggle\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\u001b[?25l\r\u001b[K     |█████▌                          | 10kB 27.7MB/s eta 0:00:01\r\u001b[K     |███████████                     | 20kB 34.6MB/s eta 0:00:01\r\u001b[K     |████████████████▋               | 30kB 39.8MB/s eta 0:00:01\r\u001b[K     |██████████████████████▏         | 40kB 30.5MB/s eta 0:00:01\r\u001b[K     |███████████████████████████▊    | 51kB 32.8MB/s eta 0:00:01\r\u001b[K     |████████████████████████████████| 61kB 8.4MB/s \n",
      "\u001b[?25h  Building wheel for kaggle (setup.py) ... \u001b[?25l\u001b[?25hdone\n"
     ]
    }
   ],
   "source": [
    "#Download kaggle cli\n",
    "! pip install --upgrade --force-reinstall --no-deps kaggle -q"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from pathlib import Path\n",
    "\n",
    "data_dir='data'\n",
    "if not Path(data_dir).exists():\n",
    "  ! mkdir -p {data_dir}\n",
    "  \n",
    "  #Download competition data\n",
    "  ! kaggle competitions download -c whale-categorization-playground\n",
    "  ! unzip -q whale-categorization-playground.zip\n",
    "\n",
    "  ! mv train/train {data_dir}/\n",
    "  ! mv test/test {data_dir}/\n",
    "  ! mv train.csv {data_dir}/\n",
    "  ! mv sample_submission.csv /\n",
    "\n",
    "  !rm -f train\n",
    "  !rm -f test"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Imports"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from nbdev.showdoc import *\n",
    "from fastai.vision.all import *\n",
    "\n",
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Setup"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "SEED=42\n",
    "set_seed(s=SEED, reproducible=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "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>Image</th>\n",
       "      <th>Id</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>00022e1a.jpg</td>\n",
       "      <td>w_e15442c</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>000466c4.jpg</td>\n",
       "      <td>w_1287fbc</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>00087b01.jpg</td>\n",
       "      <td>w_da2efe0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>001296d5.jpg</td>\n",
       "      <td>w_19e5482</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0014cfdf.jpg</td>\n",
       "      <td>w_f22f3e3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          Image         Id\n",
       "0  00022e1a.jpg  w_e15442c\n",
       "1  000466c4.jpg  w_1287fbc\n",
       "2  00087b01.jpg  w_da2efe0\n",
       "3  001296d5.jpg  w_19e5482\n",
       "4  0014cfdf.jpg  w_f22f3e3"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_csv(f'{data_dir}/train.csv')\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "new_whale    810\n",
       "w_1287fbc     34\n",
       "w_98baff9     27\n",
       "w_7554f44     26\n",
       "w_1eafe46     23\n",
       "            ... \n",
       "w_c774326      1\n",
       "w_740dfd4      1\n",
       "w_65ec378      1\n",
       "w_0c8a724      1\n",
       "w_d26cc27      1\n",
       "Name: Id, Length: 4251, dtype: int64"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.Id.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.08223350253807106"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(df.Id == 'new_whale').mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.5222300635144672"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(df.Id.value_counts() == 1).mean()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "52% of all whales have only single image associated with them. \n",
    "\n",
    "8.2% of all images contain a new whale ie not a known whale.\n",
    "\n",
    "[TODO] There is a superb writeup on what a solution to this problem might look like [here](https://www.kaggle.com/martinpiotte/whale-recognition-model-with-score-0-78563/notebook). In general, the conversation in the Kaggle [forum](https://www.kaggle.com/c/humpback-whale-identification/discussion) also seems to have some very informative threads."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "4251"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.Id.nunique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(9850, 2)"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## DataBlock"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "SZ = 224\n",
    "BS = 64\n",
    "NUM_WORKERS = 12 \n",
    "SEED = 42"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "source=Path(data_dir)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Path('data')"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "source"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = DataBlock(\n",
    "        blocks = (ImageBlock, CategoryBlock),\n",
    "        get_x =ColReader(0, pref=source/'train'),\n",
    "        get_y=ColReader(1),\n",
    "        splitter=RandomSplitter(seed=SEED),\n",
    "        item_tfms=Resize(SZ),\n",
    "        batch_tfms=aug_transforms())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#data.summary(source=df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "dls = data.dataloaders(source=df, bs=BS, num_workers=NUM_WORKERS)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#dls.show_batch()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def map5(preds,targs):\n",
    "    if type(preds) is list:\n",
    "        return torch.cat([map5fast(p, targs, 5).view(1) for p in preds ]).mean()\n",
    "    return map5kfast(preds,targs, 5)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Modeling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Downloading: \"https://download.pytorch.org/models/resnet50-19c8e357.pth\" to /root/.cache/torch/hub/checkpoints/resnet50-19c8e357.pth\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f35da6d581be4032926c40bb21e1cdd5",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(FloatProgress(value=0.0, max=102502400.0), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "learn = cnn_learner(dls, resnet50, metrics=[accuracy, map5])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#learn.lr_find()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#learn.fine_tune(5, 1e-2, freeze_epochs=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "name": "python3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}