3-predict.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "nbpresent": {
     "id": "5418a6aa-f7fd-4a96-9306-53107c8908e4"
    }
   },
   "outputs": [],
   "source": [
    "%reload_ext autoreload\n",
    "%autoreload 2\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "nbpresent": {
     "id": "535c8196-48f5-42c2-bc98-0480a5c956b8"
    }
   },
   "outputs": [],
   "source": [
    "from fastai import *\n",
    "from fastai.vision import *\n",
    "path = Path(Config.get_key('data_path')).expanduser()/'handwritten'\n",
    "#bs = 4096\n",
    "#tfms = get_transforms(do_flip=False)\n",
    "#data = ImageDataBunch.from_folder(path, train='Train', valid='Validation', ds_tfms=tfms, size=32, bs=bs)\n",
    "#learn = create_cnn(data, models.resnet34, metrics=error_rate)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "nbpresent": {
     "id": "85eb5414-b1a6-44bf-afc4-becef4c18318"
    }
   },
   "outputs": [],
   "source": [
    "classes = ['#', '$', '&', '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', '@', 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M', 'N', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "nbpresent": {
     "id": "5631d8f9-4fdc-4c6c-ac09-b85cc304430a"
    }
   },
   "outputs": [],
   "source": [
    "#empty_data = ImageDataBunch.single_from_classes(path, classes, tfms=get_transforms()).normalize(imagenet_stats)\n",
    "empty_data = ImageDataBunch.single_from_classes(path, classes, tfms=get_transforms())\n",
    "learn = create_cnn(empty_data, models.resnet34)\n",
    "# The model was created as DataParallel, therefore we need to load it as DataParallel for inference.\n",
    "learn.model = torch.nn.DataParallel(learn.model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "nbpresent": {
     "id": "c636595e-3372-4ab6-9bd6-bccb93d4d655"
    }
   },
   "outputs": [],
   "source": [
    "learn.load('stage-2-resnet34-32px')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "nbpresent": {
     "id": "557d8821-ef3a-434a-b653-84311e122f69"
    }
   },
   "outputs": [],
   "source": [
    "img_folder = path/'Validation/J'\n",
    "img_path = path/'Validation/J/50.jpg'\n",
    "#img_folder.ls()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "nbpresent": {
     "id": "016ae93c-a4ed-4777-b178-64217dafb678"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#img = open_image(data.train_ds.x[50000])\n",
    "img = open_image(img_path)\n",
    "img.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "nbpresent": {
     "id": "236868a5-b343-4ef9-a7e4-8fee6f8dc987"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "('J',\n",
       " tensor(23),\n",
       " tensor([5.0915e-01, 6.7921e-01, 3.3920e-01, 7.6282e-02, 2.0044e+01, 1.2925e+00,\n",
       "         2.4058e+00, 2.7680e-02, 9.9432e+02, 1.4533e-01, 2.1860e+00, 1.5286e-01,\n",
       "         8.1559e-03, 1.8648e-02, 8.8276e-04, 9.5635e-01, 1.3836e-01, 1.1316e+00,\n",
       "         8.4145e-01, 3.1691e+01, 2.3780e-02, 5.1446e-02, 3.8148e+02, 6.8944e+04,\n",
       "         1.8652e+00, 5.0135e+01, 6.5295e-02, 2.4911e-02, 1.9939e-01, 2.1289e-02,\n",
       "         1.0846e+00, 2.0734e+00, 2.3412e+02, 1.2138e+00, 2.7126e-01, 1.8726e-02,\n",
       "         1.1717e+00, 9.7827e-01, 1.9378e+01]))"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "learn.predict(img)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "nbpresent": {
     "id": "ecc9b4ac-f7bf-49a6-b4cb-5c10271cd363"
    }
   },
   "outputs": [],
   "source": [
    "#! wget http://yourstorage.blob.core.windows.net/ficheros/e.png"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "img = open_image(\"e.png\")\n",
    "img.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "('T',\n",
       " tensor(32),\n",
       " tensor([1.0580e+00, 6.7378e-01, 1.1900e+00, 3.0151e+01, 1.1550e+00, 7.5903e+00,\n",
       "         5.9660e-01, 9.1403e-01, 1.1940e-01, 1.5945e-01, 2.9491e+00, 6.3246e-02,\n",
       "         7.6829e-02, 1.8674e-01, 3.2650e-01, 2.4182e+00, 6.4433e+00, 1.1289e+01,\n",
       "         9.0881e+00, 3.4065e+00, 1.4036e-02, 8.5021e-02, 1.3288e+01, 8.6918e+00,\n",
       "         9.4362e-01, 3.1971e+00, 6.2158e-01, 2.2171e+00, 3.5839e+01, 2.2274e-01,\n",
       "         5.9771e+00, 2.1479e-01, 8.9982e+01, 7.7744e-01, 4.9226e+00, 1.0818e+01,\n",
       "         7.8822e-01, 1.1060e-01, 5.6203e+00]))"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "learn.predict(img)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "#! rm 4.png*\n",
    "#! wget http://yourstorage.blob.core.windows.net/ficheros/4.png"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 216x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "img = open_image(\"4.png\")\n",
    "img.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "('2',\n",
       " tensor(5),\n",
       " tensor([2.8651e+01, 7.8163e-01, 2.4273e+00, 9.3279e-04, 3.9651e+00, 4.6861e+02,\n",
       "         2.9150e+00, 3.0312e-01, 8.4479e-01, 1.6170e-02, 2.0840e-01, 1.9164e-03,\n",
       "         1.8348e+00, 3.2764e-01, 3.2780e+02, 3.8034e+00, 3.2383e-01, 2.8538e+00,\n",
       "         3.8291e+00, 1.5664e-02, 2.5684e+00, 2.4984e+00, 6.2430e+00, 5.1964e+01,\n",
       "         3.4403e-01, 8.5277e+00, 4.9288e+00, 2.1688e+01, 7.3078e-01, 5.4234e+00,\n",
       "         1.5278e+02, 2.3193e+01, 1.1537e-02, 1.5025e-01, 5.3726e+00, 5.8060e+00,\n",
       "         2.5256e-02, 8.1040e-01, 1.1062e+01]))"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "learn.predict(img)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x69d17b8>"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import cv2\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "img = cv2.imread('4.png')\n",
    "img = cv2.bitwise_not(img)\n",
    "img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)\n",
    "\n",
    "\n",
    "plt.imshow(img)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cv2.imwrite('4invert.png',img)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "learn.predict(img)"
   ]
  }
 ],
 "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"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}