peek-logits.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np # linear algebra\n",
    "import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n",
    "import libmr\n",
    "import matplotlib.pyplot as plt\n",
    "from IPython.display import Image\n",
    "from IPython.core.display import HTML \n",
    "from PIL import Image\n",
    "import seaborn as sns\n",
    "from PIL.ExifTags import TAGS, GPSTAGS\n",
    "from urllib import request\n",
    "from io import BytesIO\n",
    "%matplotlib inline "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# npz files are generated when running train.py with either -t or -tt\n",
    "landmark_logits = np.load('logits/6696.npz')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "46\n"
     ]
    }
   ],
   "source": [
    "# they contain a dictionary with landmarks -> np.array of 14951 items (logit values)\n",
    "n_ids = len(landmark_logits.keys())\n",
    "print(n_ids)\n",
    "c_cats = 14951"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# pick top 2 predicted landmarks and store its logits for each prediction\n",
    "# note though some predictions may be wrong\n",
    "logits = np.empty((n_ids,2))\n",
    "for i, (k,_logits) in enumerate(landmark_logits.items()):\n",
    "    order = np.argsort(_logits)\n",
    "    logits[i,0] = _logits[order[-1]]\n",
    "    logits[i,1] = _logits[order[-2]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0],\n",
       "       [1, 0]])"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.argsort(logits)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1d0d08eda0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize = (8, 8))\n",
    "plt.title('Logits distribuition for top 2 predicted classes')\n",
    "sns.distplot(logits[:,0],color='green', kde=True,bins=100)\n",
    "sns.distplot(logits[:,1],color='red', kde=True,bins=100)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda root]",
   "language": "python",
   "name": "conda-root-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}