Data-Exploration.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "shape for each channel.\n",
      "(352366, 32, 32, 8)\n",
      "(352366, 32, 32, 10)\n",
      "(352366, 17)\n",
      "------------------------------------------------------------\n",
      "validation part\n",
      "(24119, 32, 32, 8)\n",
      "(24119, 32, 32, 10)\n",
      "(24119, 17)\n",
      "show class distribution\n",
      "[ 5068. 24431. 31693.  8651. 16493. 35290.  3269. 39326. 13584. 11954.\n",
      " 42902.  9514.  9165. 41377.  2392.  7898. 49359.]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "Text(0.5,1,'class distribution')"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import h5py\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import os\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "base_dir = \"/home/jpbreuer/alibaba-ai/\"\n",
    "path_training = os.path.join(base_dir, \"training.h5\")\n",
    "path_test = os.path.join(base_dir, \"validation.h5\")\n",
    "\n",
    "fid_training = h5py.File(path_training,'r')\n",
    "fid_validation = h5py.File(path_test,'r')\n",
    "\n",
    "print(\"shape for each channel.\")\n",
    "s1_training = fid_training['sen1']\n",
    "print(s1_training.shape)\n",
    "s2_training = fid_training['sen2']\n",
    "print(s2_training.shape)\n",
    "label_training = fid_training['label']\n",
    "print(label_training.shape)\n",
    "\n",
    "print(\"-\" * 60)\n",
    "print(\"validation part\")\n",
    "s1_validation = fid_validation['sen1']\n",
    "print(s1_validation.shape)\n",
    "s2_validation = fid_validation['sen2']\n",
    "print(s2_validation.shape)\n",
    "label_validation = fid_validation['label']\n",
    "print(label_validation.shape)\n",
    "\n",
    "print(\"show class distribution\")\n",
    "label_qty = np.sum(fid_training[\"label\"], axis=0)\n",
    "print(label_qty)\n",
    "plt.plot(label_qty)\n",
    "plt.title(\"class distribution\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 4 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# visualization, plot the first pair of Sentinel-1 and Sentinel-2 patches of training.h5\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "plt.subplot(121)\n",
    "plt.imshow(10*np.log10(s1_training[0,:,:,4]),cmap=plt.cm.get_cmap('gray'));\n",
    "plt.colorbar()\n",
    "plt.title('Sentinel-1')\n",
    "\n",
    "plt.subplot(122)\n",
    "plt.imshow(s2_training[0,:,:,1],cmap=plt.cm.get_cmap('gray'));\n",
    "plt.colorbar()\n",
    "plt.title('Sentinel-2')\n",
    "\n",
    "plt.show()\n",
    "\n",
    "### as you can see, it is difficult to identify the image as a class by human."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/lib/python3.7/site-packages/sklearn/ensemble/weight_boosting.py:29: DeprecationWarning: numpy.core.umath_tests is an internal NumPy module and should not be imported. It will be removed in a future NumPy release.\n",
      "  from numpy.core.umath_tests import inner1d\n"
     ]
    },
    {
     "ename": "MemoryError",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mMemoryError\u001b[0m                               Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-5-a0dc1b6b7d8d>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0;31m## for a small dataset train and test\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 6\u001b[0;31m \u001b[0mtrain_s1\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0marray\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfid_training\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'sen1'\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      7\u001b[0m \u001b[0mtrain_s2\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0marray\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfid_training\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'sen2'\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      8\u001b[0m \u001b[0mtrain_label\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0marray\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfid_training\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'label'\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32mh5py/_objects.pyx\u001b[0m in \u001b[0;36mh5py._objects.with_phil.wrapper\u001b[0;34m()\u001b[0m\n",
      "\u001b[0;32mh5py/_objects.pyx\u001b[0m in \u001b[0;36mh5py._objects.with_phil.wrapper\u001b[0;34m()\u001b[0m\n",
      "\u001b[0;32m/usr/lib/python3.7/site-packages/h5py/_hl/dataset.py\u001b[0m in \u001b[0;36m__array__\u001b[0;34m(self, dtype)\u001b[0m\n\u001b[1;32m    688\u001b[0m         \u001b[0myou\u001b[0m \u001b[0mhave\u001b[0m \u001b[0mto\u001b[0m \u001b[0mread\u001b[0m \u001b[0mthe\u001b[0m \u001b[0mwhole\u001b[0m \u001b[0mdataset\u001b[0m \u001b[0mevery\u001b[0m \u001b[0mtime\u001b[0m \u001b[0mthis\u001b[0m \u001b[0mmethod\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0mcalled\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    689\u001b[0m         \"\"\"\n\u001b[0;32m--> 690\u001b[0;31m         \u001b[0marr\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mnumpy\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mempty\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshape\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mdtype\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mdtype\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0mdtype\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mNone\u001b[0m \u001b[0;32melse\u001b[0m \u001b[0mdtype\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    691\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    692\u001b[0m         \u001b[0;31m# Special case for (0,)*-shape datasets\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mMemoryError\u001b[0m: "
     ]
    }
   ],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.metrics import classification_report\n",
    "\n",
    "## for a small dataset train and test\n",
    "\n",
    "train_s1 = np.array(fid_training['sen1'])\n",
    "train_s2 = np.array(fid_training['sen2'])\n",
    "train_label = np.array(fid_training['label'])\n",
    "train_y = np.argmax(train_label, axis=1)\n",
    "\n",
    "\n",
    "validation_s1 = np.array(fid_validation['sen1'])\n",
    "validation_s2 = np.array(fid_validation['sen2'])\n",
    "validation_label = np.array(fid_validation['label'])\n",
    "validation_y = np.argmax(validation_label, axis=1)\n",
    "\n",
    "\n",
    "n = train_s1.shape[0]\n",
    "train_s1 = train_s1.reshape((n, -1))\n",
    "train_s2 = train_s2.reshape((n, -1))\n",
    "train_X = np.hstack([train_s1, train_s2])\n",
    "\n",
    "n = validation_s1.shape[0]\n",
    "validation_s1 = validation_s1.reshape((n, -1))\n",
    "validation_s2 = validation_s2.reshape((n, -1))\n",
    "validation_X = np.hstack([validation_s1, validation_s2])\n",
    "\n",
    "clf = RandomForestClassifier()\n",
    "clf.fit(train_X, train_y)\n",
    "pre_val_y = clf.predict(validation_X)\n",
    "\n",
    "print(classification_report(validation_y, pre_val_y))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "done 0/352366\n",
      "done 100/352366\n",
      "done 200/352366\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/lib/python3.7/site-packages/sklearn/linear_model/stochastic_gradient.py:128: FutureWarning: max_iter and tol parameters have been added in <class 'sklearn.linear_model.stochastic_gradient.SGDClassifier'> in 0.19. If both are left unset, they default to max_iter=5 and tol=None. If tol is not None, max_iter defaults to max_iter=1000. From 0.21, default max_iter will be 1000, and default tol will be 1e-3.\n",
      "  \"and default tol will be 1e-3.\" % type(self), FutureWarning)\n"
     ]
    }
   ],
   "source": [
    "### simple classification example\n",
    "### Training part using batch\n",
    "\n",
    "from sklearn.linear_model import SGDClassifier\n",
    "\n",
    "train_s1 = s1_training\n",
    "train_s2 = s2_training\n",
    "train_label = label_training\n",
    "clf = SGDClassifier()\n",
    "\n",
    "train_y = np.argmax(train_label, axis=1)\n",
    "classes = list(set(train_y))\n",
    "batch_size = 100\n",
    "n_sampels = train_s1.shape[0]\n",
    "\n",
    "for i in range(0, n_sampels, batch_size):\n",
    "    ## this is an idea for batch training\n",
    "    ## you can relpace this loop for deep learning methods\n",
    "    if i % batch_size * 10 == 0:\n",
    "        print(\"done %d/%d\" % (i, n_sampels))\n",
    "    start_pos = i\n",
    "    end_pos = min(i + batch_size, n_sampels)\n",
    "    train_s1_batch = np.asarray(train_s1[start_pos:end_pos, :, :, :])\n",
    "    train_s2_batch = np.asarray(train_s2[start_pos:end_pos, :, :, :])\n",
    "    cur_batch_size = train_s2_batch.shape[0]\n",
    "    train_s1_batch = train_s1_batch.reshape((cur_batch_size, -1))\n",
    "    train_s2_batch = train_s2_batch.reshape((cur_batch_size, -1))\n",
    "    train_X_batch = np.hstack([train_s1_batch, train_s2_batch])\n",
    "    label_batch = train_y[start_pos:end_pos]\n",
    "    clf.partial_fit(train_X_batch, label_batch, classes=classes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "### make a prediction on validation\n",
    "pred_y = []\n",
    "train_val_y = np.argmax(label_validation, axis=1)\n",
    "batch_size = 100\n",
    "n_val_samples = s2_validation.shape[0]\n",
    "for i in range(0, n_val_samples, batch_size):\n",
    "    start_pos = i\n",
    "    end_pos = min(i + batch_size, n_val_samples)\n",
    "    val_s1_batch = np.asarray(s1_validation[start_pos:end_pos, :, :, :])\n",
    "    val_s2_batch = np.asarray(s2_validation[start_pos:end_pos, :, :, :])\n",
    "    cur_batch_size = val_s2_batch.shape[0]\n",
    "    val_s1_batch = val_s1_batch.reshape((cur_batch_size, -1))\n",
    "    val_s2_batch = val_s2_batch.reshape((cur_batch_size, -1))\n",
    "    val_X_batch = np.hstack([val_s1_batch, val_s2_batch])\n",
    "    tmp_pred_y = clf.predict(val_X_batch)\n",
    "    pred_y.append(tmp_pred_y)\n",
    "pred_y = np.hstack(pred_y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import classification_report\n",
    "print(classification_report(train_val_y, pred_y))"
   ]
  }
 ],
 "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": "3.7.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}