bharadwaj6/animal-recognition.ipynb

## animal-recognition.ipynb
{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Multi-label classification"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "%reload_ext autoreload\n",
    "%autoreload 2\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from fastai.conv_learner import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "PATH = 'data/animal-recognition/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Data preparation steps if you are using Crestle:\n",
    "\n",
    "os.makedirs('data/animal-recognition/models', exist_ok=True)\n",
    "os.makedirs('cache/animal-recognition/tmp', exist_ok=True)\n",
    "\n",
    "# !ln -s /datasets/kaggle/planet-understanding-the-amazon-from-space/train-jpg {PATH}\n",
    "# !ln -s /datasets/kaggle/planet-understanding-the-amazon-from-space/test-jpg {PATH}\n",
    "# !ln -s /datasets/kaggle/planet-understanding-the-amazon-from-space/train_v2.csv {PATH}\n",
    "# !ln -s cache/planet/tmp {PATH}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "attributes.txt  \u001b[0m\u001b[01;34mmodels\u001b[0m/  \u001b[01;34msubmissions\u001b[0m/  \u001b[01;34mtmp\u001b[0m/\r\n",
      "\u001b[01;34mmeta-data\u001b[0m/      preload  \u001b[01;34mtest_img\u001b[0m/     \u001b[01;34mtrain_img\u001b[0m/\r\n"
     ]
    }
   ],
   "source": [
    "ls {PATH}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "heading_collapsed": true
   },
   "source": [
    "## Multi-label versus single-label classification"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "hidden": true
   },
   "outputs": [],
   "source": [
    "from fastai.plots import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "hidden": true
   },
   "outputs": [],
   "source": [
    "def get_1st(path): return glob(f'{path}/*.*')[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "hidden": true
   },
   "source": [
    "In multi-label classification each sample can belong to one or more clases. In the previous example, the first images belongs to two clases: *haze* and *primary*. The second image belongs to four clases: *agriculture*, *clear*, *primary* and  *water*."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Multi-label models for Planet dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "from planet import f2\n",
    "\n",
    "metrics=[f2]\n",
    "f_model = resnet50\n",
    "bs = 128"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# def convert_csv(input_file, output_file):\n",
    "#     nlines = []\n",
    "#     index = None\n",
    "#     attrs = None\n",
    "#     with open(input_file, 'r') as f:\n",
    "#         lines = f.readlines()\n",
    "#         index = 'Image_name,tag'\n",
    "#         attrs = lines[0].strip().split(',')[1:]\n",
    "#         for line in lines[1:]:\n",
    "#             line = line.strip().split(',')\n",
    "#             nums = line[1:]\n",
    "#             tags = []\n",
    "#             for i in range(len(nums)):\n",
    "#                 if nums[i] == '1':\n",
    "#                     tags.append(attrs[i])\n",
    "#             nlines.append(line[0] + ',' + ' '.join(tags))\n",
    "#     with open(output_file, 'w+') as of:\n",
    "#         out = index + '\\n' + '\\n'.join(nlines)\n",
    "#         of.write(out)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# convert_csv(f'{PATH}/meta-data/train.csv', f'{PATH}/meta-data/train_labels.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "labelcsv = f'{PATH}/meta-data/train_labels.csv'\n",
    "n = len(list(open(labelcsv)))-1\n",
    "val_idxs = get_cv_idxs(n)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "transforms = transforms_side_on + [RandomDihedral()]\n",
    "def get_data(sz):\n",
    "    tfms = tfms_from_model(f_model, sz, aug_tfms=transforms, max_zoom=1.05)\n",
    "    return ImageClassifierData.from_csv(PATH, 'train_img', labelcsv, tfms=tfms,\n",
    "                    val_idxs=val_idxs, test_name='test_img', bs=bs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = get_data(256)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "x,y = next(iter(data.val_dl))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['attrib_01',\n",
       " 'attrib_02',\n",
       " 'attrib_03',\n",
       " 'attrib_04',\n",
       " 'attrib_05',\n",
       " 'attrib_06',\n",
       " 'attrib_07',\n",
       " 'attrib_08',\n",
       " 'attrib_09',\n",
       " 'attrib_10',\n",
       " 'attrib_11',\n",
       " 'attrib_12',\n",
       " 'attrib_13',\n",
       " 'attrib_14',\n",
       " 'attrib_15',\n",
       " 'attrib_16',\n",
       " 'attrib_17',\n",
       " 'attrib_18',\n",
       " 'attrib_19',\n",
       " 'attrib_20',\n",
       " 'attrib_21',\n",
       " 'attrib_22',\n",
       " 'attrib_23',\n",
       " 'attrib_24',\n",
       " 'attrib_25',\n",
       " 'attrib_26',\n",
       " 'attrib_27',\n",
       " 'attrib_28',\n",
       " 'attrib_29',\n",
       " 'attrib_30',\n",
       " 'attrib_31',\n",
       " 'attrib_32',\n",
       " 'attrib_33',\n",
       " 'attrib_34',\n",
       " 'attrib_35',\n",
       " 'attrib_36',\n",
       " 'attrib_37',\n",
       " 'attrib_38',\n",
       " 'attrib_39',\n",
       " 'attrib_40',\n",
       " 'attrib_41',\n",
       " 'attrib_42',\n",
       " 'attrib_43',\n",
       " 'attrib_44',\n",
       " 'attrib_45',\n",
       " 'attrib_46',\n",
       " 'attrib_47',\n",
       " 'attrib_48',\n",
       " 'attrib_49',\n",
       " 'attrib_50',\n",
       " 'attrib_51',\n",
       " 'attrib_52',\n",
       " 'attrib_53',\n",
       " 'attrib_54',\n",
       " 'attrib_55',\n",
       " 'attrib_56',\n",
       " 'attrib_57',\n",
       " 'attrib_58',\n",
       " 'attrib_59',\n",
       " 'attrib_60',\n",
       " 'attrib_61',\n",
       " 'attrib_62',\n",
       " 'attrib_63',\n",
       " 'attrib_64',\n",
       " 'attrib_65',\n",
       " 'attrib_66',\n",
       " 'attrib_67',\n",
       " 'attrib_68',\n",
       " 'attrib_69',\n",
       " 'attrib_70',\n",
       " 'attrib_71',\n",
       " 'attrib_72',\n",
       " 'attrib_73',\n",
       " 'attrib_74',\n",
       " 'attrib_75',\n",
       " 'attrib_76',\n",
       " 'attrib_77',\n",
       " 'attrib_78',\n",
       " 'attrib_79',\n",
       " 'attrib_80',\n",
       " 'attrib_81',\n",
       " 'attrib_82',\n",
       " 'attrib_83',\n",
       " 'attrib_84',\n",
       " 'attrib_85']"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.classes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "\n",
       "    0     0     0  ...      0     0     1\n",
       "    1     0     0  ...      1     0     0\n",
       "    1     0     0  ...      1     0     0\n",
       "       ...          ⋱          ...       \n",
       "    1     0     0  ...      0     0     1\n",
       "    0     0     0  ...      0     0     0\n",
       "    1     0     0  ...      1     0     0\n",
       "[torch.cuda.FloatTensor of size 128x85 (GPU 0)]"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('attrib_01', 0.0),\n",
       " ('attrib_02', 0.0),\n",
       " ('attrib_03', 0.0),\n",
       " ('attrib_04', 0.0),\n",
       " ('attrib_05', 0.0),\n",
       " ('attrib_06', 1.0),\n",
       " ('attrib_07', 1.0),\n",
       " ('attrib_08', 0.0),\n",
       " ('attrib_09', 0.0),\n",
       " ('attrib_10', 0.0),\n",
       " ('attrib_11', 0.0),\n",
       " ('attrib_12', 0.0),\n",
       " ('attrib_13', 0.0),\n",
       " ('attrib_14', 0.0),\n",
       " ('attrib_15', 0.0),\n",
       " ('attrib_16', 1.0),\n",
       " ('attrib_17', 1.0),\n",
       " ('attrib_18', 1.0),\n",
       " ('attrib_19', 0.0),\n",
       " ('attrib_20', 0.0),\n",
       " ('attrib_21', 1.0),\n",
       " ('attrib_22', 0.0),\n",
       " ('attrib_23', 1.0),\n",
       " ('attrib_24', 0.0),\n",
       " ('attrib_25', 0.0),\n",
       " ('attrib_26', 1.0),\n",
       " ('attrib_27', 0.0),\n",
       " ('attrib_28', 1.0),\n",
       " ('attrib_29', 0.0),\n",
       " ('attrib_30', 0.0),\n",
       " ('attrib_31', 1.0),\n",
       " ('attrib_32', 1.0),\n",
       " ('attrib_33', 1.0),\n",
       " ('attrib_34', 1.0),\n",
       " ('attrib_35', 0.0),\n",
       " ('attrib_36', 0.0),\n",
       " ('attrib_37', 0.0),\n",
       " ('attrib_38', 0.0),\n",
       " ('attrib_39', 0.0),\n",
       " ('attrib_40', 0.0),\n",
       " ('attrib_41', 1.0),\n",
       " ('attrib_42', 0.0),\n",
       " ('attrib_43', 0.0),\n",
       " ('attrib_44', 0.0),\n",
       " ('attrib_45', 1.0),\n",
       " ('attrib_46', 0.0),\n",
       " ('attrib_47', 0.0),\n",
       " ('attrib_48', 0.0),\n",
       " ('attrib_49', 0.0),\n",
       " ('attrib_50', 0.0),\n",
       " ('attrib_51', 0.0),\n",
       " ('attrib_52', 0.0),\n",
       " ('attrib_53', 0.0),\n",
       " ('attrib_54', 1.0),\n",
       " ('attrib_55', 0.0),\n",
       " ('attrib_56', 0.0),\n",
       " ('attrib_57', 0.0),\n",
       " ('attrib_58', 1.0),\n",
       " ('attrib_59', 0.0),\n",
       " ('attrib_60', 1.0),\n",
       " ('attrib_61', 0.0),\n",
       " ('attrib_62', 1.0),\n",
       " ('attrib_63', 1.0),\n",
       " ('attrib_64', 0.0),\n",
       " ('attrib_65', 0.0),\n",
       " ('attrib_66', 0.0),\n",
       " ('attrib_67', 0.0),\n",
       " ('attrib_68', 0.0),\n",
       " ('attrib_69', 0.0),\n",
       " ('attrib_70', 0.0),\n",
       " ('attrib_71', 0.0),\n",
       " ('attrib_72', 0.0),\n",
       " ('attrib_73', 0.0),\n",
       " ('attrib_74', 1.0),\n",
       " ('attrib_75', 0.0),\n",
       " ('attrib_76', 1.0),\n",
       " ('attrib_77', 0.0),\n",
       " ('attrib_78', 0.0),\n",
       " ('attrib_79', 0.0),\n",
       " ('attrib_80', 0.0),\n",
       " ('attrib_81', 0.0),\n",
       " ('attrib_82', 0.0),\n",
       " ('attrib_83', 0.0),\n",
       " ('attrib_84', 0.0),\n",
       " ('attrib_85', 1.0)]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "list(zip(data.classes, y[0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(data.val_ds.denorm(to_np(x))[0]*1.4);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "IMAGE_NAME = 'Image_name'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "sz = 64"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = get_data(sz)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "be377d154473482c826d9e0e6baf0c3e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=6), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "data = data.resize(int(sz*1.3), 'tmp')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn = ConvLearner.pretrained(f_model, data, metrics=metrics)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7dc8740dff754ec9a27910a4dd699f08",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=1), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch      trn_loss   val_loss   f2                        \n",
      "    0      0.698731   0.476298   0.82276   \n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "lrf=learn.lr_find()\n",
    "learn.sched.plot()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "lr = 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "28848776146348cd9f20585b25917e84",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=7), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch      trn_loss   val_loss   f2                        \n",
      "    0      0.446306   0.330137   0.861782  \n",
      "    1      0.400655   0.302515   0.873975                  \n",
      "    2      0.381341   0.297806   0.875669                  \n",
      "    3      0.376521   0.290617   0.879058                  \n",
      "    4      0.367477   0.282292   0.882069                  \n",
      "    5      0.362307   0.278151   0.884482                  \n",
      "    6      0.357316   0.277647   0.885311                  \n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[array([0.27765]), 0.8853110586787702]"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "learn.fit(lr, 3, cycle_len=1, cycle_mult=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.save(f'{sz}-freeze')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "9534"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import gc\n",
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.unfreeze()\n",
    "lrs = [lr/30, lr/10, lr]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3747c97825244128b229f135ac4768c1",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=7), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch      trn_loss   val_loss   f2                        \n",
      "    0      0.353853   0.28389    0.881634  \n",
      "    1      0.31627    0.259146   0.892252                  \n",
      "    2      0.277855   0.23908    0.900887                  \n",
      "    3      0.279573   0.276262   0.887302                  \n",
      "    4      0.261233   0.235352   0.902872                  \n",
      "    5      0.230056   0.212314   0.913425                  \n",
      "    6      0.208827   0.206768   0.915861                  \n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[array([0.20677]), 0.9158613295540323]"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "learn.fit(lrs, 3, cycle_len=1, cycle_mult=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.save(f'{sz}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "learn.sched.plot_loss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "sz = 128"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5e2537e3014b45df959fe62e195bbaad",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=7), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch      trn_loss   val_loss   f2                        \n",
      "    0      0.198207   0.175919   0.930414  \n",
      "    1      0.189821   0.163851   0.934975                  \n",
      "    2      0.18323    0.161126   0.936104                  \n",
      "    3      0.180901   0.156394   0.938147                  \n",
      "    4      0.175843   0.152405   0.93998                   \n",
      "    5      0.170486   0.15092    0.940661                  \n",
      "    6      0.167664   0.150204   0.94105                   \n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[array([0.1502]), 0.9410504178164384]"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "learn.set_data(get_data(sz))\n",
    "learn.freeze()\n",
    "learn.fit(lr, 3, cycle_len=1, cycle_mult=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2339"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "9f5b29bdbe9042bd85601cd263ad8021",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=7), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch      trn_loss   val_loss   f2                        \n",
      "    0      0.170244   0.121172   0.952107  \n",
      "    1      0.163445   0.135828   0.946075                  \n",
      "    2      0.126801   0.10371    0.959678                  \n",
      "    3      0.137012   0.14888    0.940103                  \n",
      "    4      0.121868   0.108715   0.957676                  \n",
      "    5      0.095862   0.090837   0.964135                   \n",
      "    6      0.080975   0.088674   0.965387                   \n",
      "\n"
     ]
    }
   ],
   "source": [
    "learn.unfreeze()\n",
    "learn.fit(lrs, 3, cycle_len=1, cycle_mult=2)\n",
    "learn.save(f'{sz}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "sz = 196"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "87e9b4a2daec43ed9af8a9dc83a0284b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=7), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch      trn_loss   val_loss   f2                         \n",
      "    0      0.073535   0.076465   0.970464  \n",
      "    1      0.072098   0.07439    0.971121                   \n",
      "    2      0.068229   0.073546   0.97168                    \n",
      "    3      0.066447   0.074008   0.971387                   \n",
      "    4      0.065762   0.072018   0.971927                   \n",
      "    5      0.063137   0.071732   0.972438                   \n",
      "    6      0.060719   0.071725   0.972148                   \n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[array([0.07172]), 0.9721476974139234]"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "learn.set_data(get_data(sz))\n",
    "learn.freeze()\n",
    "learn.fit(lr, 3, cycle_len=1, cycle_mult=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "30"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "bb46c82bee074025b3533d09f3c98a71",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=7), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch      trn_loss   val_loss   f2                         \n",
      "    0      0.095019   0.08149    0.968552  \n",
      "    1      0.090028   0.088598   0.967067                   \n",
      "    2      0.066305   0.06819    0.974038                   \n",
      "    3      0.080955   0.134801   0.9495                     \n",
      "    5      0.051119   0.063423   0.976147                   \n",
      "    6      0.040045   0.060523   0.977641                   \n",
      "\n"
     ]
    }
   ],
   "source": [
    "learn.unfreeze()\n",
    "learn.fit(lrs, 3, cycle_len=1, cycle_mult=2)\n",
    "learn.save(f'{sz}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "sz = 256"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0a6593c778984da3a2c363551bf0086c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=7), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch      trn_loss   val_loss   f2                         \n",
      "    0      0.036559   0.056188   0.979302  \n",
      "    1      0.034259   0.055997   0.979893                   \n",
      "    2      0.032492   0.055605   0.979859                   \n",
      "    3      0.032417   0.055686   0.979915                   \n",
      "    4      0.031896   0.055315   0.98016                    \n",
      "    5      0.030672   0.055215   0.980207                   \n",
      "    6      0.029629   0.055228   0.980194                   \n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[array([0.05523]), 0.9801941212334008]"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "learn.set_data(get_data(sz))\n",
    "learn.freeze()\n",
    "learn.fit(lr, 3, cycle_len=1, cycle_mult=2)\n",
    "learn.save(f'{sz}-freeze')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "30"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ca8e65e75804453182905d4b72321109",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=7), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch      trn_loss   val_loss   f2                         \n",
      "    0      0.049474   0.062348   0.977229  \n",
      "    1      0.060518   0.075004   0.97278                    \n",
      "    2      0.044404   0.05798    0.979428                   \n",
      "    3      0.057227   0.153245   0.946635                   \n",
      "    4      0.052528   0.06584    0.976325                   \n",
      "    5      0.035576   0.057293   0.979528                   \n",
      "    6      0.026104   0.053948   0.98108                    \n",
      "\n"
     ]
    }
   ],
   "source": [
    "learn.unfreeze()\n",
    "learn.fit(lrs, 3, cycle_len=1, cycle_mult=2)\n",
    "learn.save(f'{sz}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "acee9bfc1466434cbf3bb77373a41899",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Epoch', max=15), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "epoch      trn_loss   val_loss   f2                         \n",
      "    0      0.028139   0.056662   0.979625  \n",
      "    1      0.032434   0.071013   0.975282                   \n",
      "    2      0.024377   0.055842   0.98087                    \n",
      "    3      0.035064   0.086212   0.968816                   \n",
      "    4      0.035426   0.074913   0.97364                    \n",
      "    5      0.024191   0.056595   0.979917                   \n",
      "    6      0.017271   0.055424   0.980329                   \n",
      "    7      0.023222   0.091443   0.967561                   \n",
      "    8      0.035308   0.093691   0.96679                    \n",
      "    9      0.034892   0.086996   0.970386                   \n",
      "    10     0.02329    0.069517   0.976739                   \n",
      "    11     0.015719   0.060105   0.979841                   \n",
      "    12     0.011975   0.059377   0.9804                     \n",
      "    13     0.010468   0.058148   0.980797                   \n",
      "    14     0.009437   0.05746    0.981014                    \n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[array([0.05746]), 0.9810136915431772]"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# MOAR training!\n",
    "learn.fit(lrs, 4, cycle_len=1, cycle_mult=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.save(f'{sz}-extra-train')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                                             \r"
     ]
    }
   ],
   "source": [
    "multi_preds, y = learn.TTA(is_test=True)\n",
    "preds = np.mean(multi_preds, 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "THRESHOLD = 0.5\n",
    "\n",
    "def preds_to_submission_file(preds, filename):\n",
    "    print(len(preds))\n",
    "    for i in range(len(preds)):\n",
    "        preds[i] = list(map(lambda x: 1 if x > THRESHOLD else 0, preds[i]))\n",
    "    \n",
    "    sub = pd.DataFrame(preds, columns=data.classes, dtype='int32')\n",
    "    sub.insert(0, IMAGE_NAME, data.test_ds.fnames)\n",
    "    sub[IMAGE_NAME] = sub[IMAGE_NAME].apply(lambda x: x.replace('test_img/', ''))\n",
    "    sub['sort_col'] = sub[IMAGE_NAME].apply(lambda x: int(x.replace('Image-', '').replace('.jpg', ''))).astype(int)\n",
    "    sub.sort_values(by=['sort_col'], inplace=True, ascending=True)\n",
    "    sub.drop(['sort_col'], axis=1, inplace=True)\n",
    "    \n",
    "    sub.to_csv('data/animal-recognition/submissions/' + filename, index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5400\n"
     ]
    }
   ],
   "source": [
    "preds_to_submission_file(preds, 'tta_submission_extra50.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1331"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "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": "3.6.5"
  },
  "toc": {
   "colors": {
    "hover_highlight": "#DAA520",
    "navigate_num": "#000000",
    "navigate_text": "#333333",
    "running_highlight": "#FF0000",
    "selected_highlight": "#FFD700",
    "sidebar_border": "#EEEEEE",
    "wrapper_background": "#FFFFFF"
   },
   "moveMenuLeft": true,
   "nav_menu": {
    "height": "99px",
    "width": "252px"
   },
   "navigate_menu": true,
   "number_sections": true,
   "sideBar": true,
   "threshold": 4,
   "toc_cell": false,
   "toc_section_display": "block",
   "toc_window_display": false,
   "widenNotebook": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}