Diabetic_retinopathy_detection (2).ipynb

{"nbformat":4,"nbformat_minor":0,"metadata":{"colab":{"name":"Diabetic_retinopathy_detection.ipynb","provenance":[],"collapsed_sections":[],"machine_shape":"hm"},"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":"3.6.1"}},"cells":[{"cell_type":"code","metadata":{"id":"h7AlOKSMF6-j","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1606270559288,"user_tz":360,"elapsed":430,"user":{"displayName":"Gerry","photoUrl":"https://lh4.googleusercontent.com/-oUPavxcTH4Q/AAAAAAAAAAI/AAAAAAAAAG4/Edgkb0AtMhI/s64/photo.jpg","userId":"11854613171556017399"}},"outputId":"86b07060-0c59-4edd-e7f9-061d9815926d"},"source":["from google.colab import drive\n","drive.mount('/content/drive/')"],"execution_count":16,"outputs":[{"output_type":"stream","text":["Drive already mounted at /content/drive/; to attempt to forcibly remount, call drive.mount(\"/content/drive/\", force_remount=True).\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"id":"oK3UmyUqxEt-","executionInfo":{"status":"ok","timestamp":1606269067112,"user_tz":360,"elapsed":779,"user":{"displayName":"Gerry","photoUrl":"https://lh4.googleusercontent.com/-oUPavxcTH4Q/AAAAAAAAAAI/AAAAAAAAAG4/Edgkb0AtMhI/s64/photo.jpg","userId":"11854613171556017399"}}},"source":["import scipy\n","from numpy import exp\n","from matplotlib import pyplot as plt\n","import numpy as np\n","import cv2\n","import pandas as pd\n","from math import pi, sqrt\n","import pywt\n","import os\n","from sklearn.metrics import accuracy_score\n","from sklearn.svm import SVC\n","from sklearn.model_selection import train_test_split\n","from joblib import dump, load"],"execution_count":15,"outputs":[]},{"cell_type":"code","metadata":{"id":"Roj-3gVKQ48z","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1606270633488,"user_tz":360,"elapsed":450,"user":{"displayName":"Gerry","photoUrl":"https://lh4.googleusercontent.com/-oUPavxcTH4Q/AAAAAAAAAAI/AAAAAAAAAG4/Edgkb0AtMhI/s64/photo.jpg","userId":"11854613171556017399"}},"outputId":"4bc1475d-64d0-4308-e10d-c71f2da6d506"},"source":["imagecsv = pd.read_csv(r'/content/drive/MyDrive/Diabetic Retinopathy Detection/trainLabels.csv')[:3000]\n","img_pt = r'/content/drive/MyDrive/Diabetic Retinopathy Detection/sample_images_resized'\n","imagecsv.head"],"execution_count":17,"outputs":[{"output_type":"execute_result","data":{"text/plain":["<bound method NDFrame.head of            image  level\n","0        10_left      0\n","1       10_right      0\n","2        13_left      0\n","3       13_right      0\n","4        15_left      1\n","...          ...    ...\n","2995  3706_right      0\n","2996   3708_left      0\n","2997  3708_right      0\n","2998   3710_left      0\n","2999  3710_right      0\n","\n","[3000 rows x 2 columns]>"]},"metadata":{"tags":[]},"execution_count":17}]},{"cell_type":"code","metadata":{"colab":{"base_uri":"https://localhost:8080/","height":526},"id":"FrrJU1A4Hjeg","executionInfo":{"status":"ok","timestamp":1606270642407,"user_tz":360,"elapsed":981,"user":{"displayName":"Gerry","photoUrl":"https://lh4.googleusercontent.com/-oUPavxcTH4Q/AAAAAAAAAAI/AAAAAAAAAG4/Edgkb0AtMhI/s64/photo.jpg","userId":"11854613171556017399"}},"outputId":"e43f21d4-8928-417d-8801-466bb6cdf466"},"source":["ax = imagecsv['level'].value_counts().plot(kind='bar',\n","                                    figsize=(14,8),\n","                                    title=\"ImageCsv Data\")\n","ax.set_xlabel(\"Level\")\n","ax.set_ylabel(\"Images\")"],"execution_count":18,"outputs":[{"output_type":"execute_result","data":{"text/plain":["Text(0, 0.5, 'Images')"]},"metadata":{"tags":[]},"execution_count":18},{"output_type":"display_data","data":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAA0kAAAHsCAYAAAAHN+Y0AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAcL0lEQVR4nO3df7DldX3f8dcb8DfKD1kZBOKShNRBrUQ3oOmPaFXkR3RxxqhohBI7NIka0yRNaDMtxkSH6TR1JFVTUlBMVUQbFZUKhKj50ZiwEIqiJqwKBUTYBOVHMCj47h/ne9NP1l246D337N77eMzcued8vt9zzvsuZ4Dnfr/ne6u7AwAAwMweix4AAABgVyKSAAAABiIJAABgIJIAAAAGIgkAAGAgkgAAAAYiCQAAYCCSAHjQquq6qnruoufYXlU9tKpeX1XXVtXfTnOeW1UbV/A1/mVV3VdVd01fX66qd1TVDz2I53hnVf3GSs0EwMoSSQCsJR9I8sIkL0+yT5KnJrkiyXNW+HX+tLv3nl7juUm+keSKqnryCr8OAAsgkgD4nkxHVv6kqt5cVV+vqi9V1Y9O6zdU1a1Vdcqw/wlV9RdVdce0/fXbPd/JVXV9Vf1NVf2H8ahVVe1RVadX1Ren7RdU1f7TtucmeV6Szd19eXff2923d/dbu/ucYdYvVdWd0xGgV1TVw6a5nzzMsKGqvlFVj7u/n7277+vuL3b3zyb5VJK//1mq6v1V9dWqur2q/rCqnjStn5bkFUl+eToS9ZFpfennurOqPldVL/ru/6kA8L0QSQCshKOTXJ3ksUnek+T8JD+S5AeT/GSS/1pVe0/7/m2Sk5Psm+SEJD9TVScmSVUdkeRtmUXEQZkdqTl4eJ3XJjkxyY8leXySryV567TtuUn+vLtv2NGAVfWoJGclOa67H53kR5Nc1d33JPm9JCcNu78kyae6+9YH8Wfwe0n+2XD/fyU5PMnjklyZ5N1J0t1nT7f/U3fv3d0vmPb/4vT4fZL8WpL/UVUHPYjXB2CFiCQAVsKXu/sd3X1fkvclOTTJG7r7nu6+JMk3MwumdPcnu/sz3f3t7r46yXszi54keXGSj3T3H3f3N5P8xyQ9vM5PJ/nV7r5xipvXJ3lxVe2VWaDd/ABzfjvJk6vqEd19c3dfM62/J8nLhv1ePq09GF9Jsv/Sne4+t7vvHOZ8alXts7MHd/f7u/sr05/L+5Jcm+SoBzkDACtAJAGwEm4Zbn8jSbp7+7W9k6Sqjq6qT1TVtqq6PbPwOWDa7/FJ/v5IUHffneRvhud5QpIPTqfHfT3J55Pcl+TAab+dHnnp7r9N8tLp9W6uqo9V1ROnzZ9I8shpto1JjkzyweX/+ElmR7xum37GPavqzOn0uTuSXDftc8DOHjydZnjV8LM9+f72B2B+RBIAq+09SS5Mcmh375Pkt5PUtO3mJIcs7VhVj8jsCNGSGzI7XW7f4evh3X1Tkt9PclRVHZKd6O6Lu/t5mcXUF5L8zrR+X5ILMjvl7qQkH+3uOx/kz/WiJH803X55ks2ZnQK4T5KNSz/S0ijjA6vqCdMsr0ny2O7eN8lnh/0BWEUiCYDV9ugkt3X331XVUZkFxZIPJHnBdOGHh2Z2mtoYCr+d5I1TVCxdYGFzknT37ye5NLMjTU+vqr2q6tFV9dNV9VNVdWBVbZ4+m3RPkrsyO/1uyXsyO9L0iizzVLvpiNFhVfVbSZ6V2WeJln7GezI7uvXIJG/a7qG3JPn+4f6jMgunbdPznprZkSQAFkAkAbDafjbJG6rqzsw+c3TB0obpM0KvzezCDzdnFjK3ZhYcSfKWzI5CXTI9/tOZXTRiyYuTXJTZ56Juz+xozKbMjjLtkeQXMvvs0G2ZfQ7qZ4bX/rPMLirx+MwuunB/nllVdyW5I8knkzwmyY9092em7e9Kcn2Sm5J8bppzdE6SI6ZT6z7U3Z9L8ptJ/jSzgHpKkj95gBkAmJPq7gfeCwAWYLoi3teTHN7dX170PACsD44kAbBLqaoXVNUjp9Pi/nOSz+T/X/gAAOZOJAGwq9mc2SlxX8ns9wy9rJ32AMAqcrodAADAwJEkAACAgUgCAAAY7LXoAebhgAMO6I0bNy56DAAAYBd2xRVX/HV3b9h+fU1G0saNG7Nly5ZFjwEAAOzCqur6Ha073Q4AAGAgkgAAAAYiCQAAYCCSAAAABiIJAABgIJIAAAAGIgkAAGAgkgAAAAYiCQAAYCCSAAAABiIJAABgIJIAAAAGIgkAAGAgkgAAAAYiCQAAYCCSAAAABiIJAABgIJIAAAAGey16gPVo4+kfW/QIu6Xrzjxh0SMAALAOOJIEAAAwEEkAAAADkQQAADAQSQAAAAORBAAAMBBJAAAAA5EEAAAwEEkAAAADkQQAADAQSQAAAAORBAAAMBBJAAAAA5EEAAAwEEkAAAADkQQAADAQSQAAAAORBAAAMBBJAAAAA5EEAAAwEEkAAAADkQQAADAQSQAAAAORBAAAMBBJAAAAA5EEAAAwEEkAAAADkQQAADAQSQAAAAORBAAAMBBJAAAAA5EEAAAwEEkAAAADkQQAADAQSQAAAAORBAAAMBBJAAAAA5EEAAAwEEkAAAADkQQAADAQSQAAAAORBAAAMBBJAAAAg7lFUlUdWlWfqKrPVdU1VfW6aX3/qrq0qq6dvu83rVdVnVVVW6vq6qp62vBcp0z7X1tVp8xrZgAAgHkeSbo3yS929xFJnpHk1VV1RJLTk1zW3YcnuWy6nyTHJTl8+jotyduTWVQlOSPJ0UmOSnLGUlgBAACstLlFUnff3N1XTrfvTPL5JAcn2ZzkvGm385KcON3enORdPfPpJPtW1UFJnp/k0u6+rbu/luTSJMfOa24AAGB9W5XPJFXVxiQ/nOTPkhzY3TdPm76a5MDp9sFJbhgeduO0trP17V/jtKraUlVbtm3btqLzAwAA68fcI6mq9k7yP5P8fHffMW7r7k7SK/E63X12d2/q7k0bNmxYiacEAADWoblGUlU9JLNAend3/960fMt0Gl2m77dO6zclOXR4+CHT2s7WAQAAVtw8r25XSc5J8vnu/i/DpguTLF2h7pQkHx7WT56ucveMJLdPp+VdnOSYqtpvumDDMdMaAADAittrjs/9T5K8Mslnquqqae3fJzkzyQVV9aok1yd5ybTtoiTHJ9ma5O4kpyZJd99WVb+e5PJpvzd0921znBsAAFjH5hZJ3f3HSWonm5+zg/07yat38lznJjl35aYDAADYsVW5uh0AAMDuQiQBAAAMRBIAAMBAJAEAAAxEEgAAwEAkAQAADEQSAADAQCQBAAAMRBIAAMBAJAEAAAxEEgAAwEAkAQAADEQSAADAQCQBAAAMRBIAAMBAJAEAAAxEEgAAwEAkAQAADEQSAADAQCQBAAAMRBIAAMBAJAEAAAxEEgAAwEAkAQAADEQSAADAQCQBAAAMRBIAAMBAJAEAAAxEEgAAwEAkAQAADEQSAADAQCQBAAAMRBIAAMBAJAEAAAxEEgAAwEAkAQAADEQSAADAQCQBAAAMRBIAAMBAJAEAAAxEEgAAwEAkAQAADEQSAADAQCQBAAAMRBIAAMBAJAEAAAxEEgAAwEAkAQAADEQSAADAQCQBAAAMRBIAAMBAJAEAAAxEEgAAwEAkAQAADEQSAADAQCQBAAAMRBIAAMBAJAEAAAxEEgAAwEAkAQAADEQSAADAQCQBAAAMRBIAAMBAJAEAAAxEEgAAwEAkAQAADEQSAADAQCQBAAAMRBIAAMBAJAEAAAxEEgAAwEAkAQAADEQSAADAQCQBAAAMRBIAAMBAJAEAAAxEEgAAwEAkAQAADEQSAADAQCQBAAAMRBIAAMBAJAEAAAxEEgAAwGBukVRV51bVrVX12WHt9VV1U1VdNX0dP2z7d1W1tar+sqqeP6wfO61trarT5zUvAABAMt8jSe9McuwO1t/c3UdOXxclSVUdkeRlSZ40PeZtVbVnVe2Z5K1JjktyRJKTpn0BAADmYq95PXF3/2FVbVzm7puTnN/d9yT5clVtTXLUtG1rd38pSarq/Gnfz63wuAAAAEkW85mk11TV1dPpePtNawcnuWHY58ZpbWfr36GqTquqLVW1Zdu2bfOYGwAAWAdWO5LenuQHkhyZ5OYkv7lST9zdZ3f3pu7etGHDhpV6WgAAYJ2Z2+l2O9LdtyzdrqrfSfLR6e5NSQ4ddj1kWsv9rAMAAKy4VT2SVFUHDXdflGTpyncXJnlZVT2sqg5LcniSP09yeZLDq+qwqnpoZhd3uHA1ZwYAANaXuR1Jqqr3JnlWkgOq6sYkZyR5VlUdmaSTXJfkXydJd19TVRdkdkGGe5O8urvvm57nNUkuTrJnknO7+5p5zQwAADDPq9udtIPlc+5n/zcmeeMO1i9KctEKjgYAALBTi7i6HQAAwC5LJAEAAAxEEgAAwEAkAQAADEQSAADAQCQBAAAMRBIAAMBAJAEAAAxEEgAAwEAkAQAADEQSAADAQCQBAAAMRBIAAMBAJAEAAAxEEgAAwEAkAQAADEQSAADAQCQBAAAMRBIAAMBAJAEAAAxEEgAAwEAkAQAADEQSAADAQCQBAAAMRBIAAMBAJAEAAAxEEgAAwEAkAQAADEQSAADAQCQBAAAMRBIAAMBAJAEAAAxEEgAAwEAkAQAADEQSAADAQCQBAAAMRBIAAMBAJAEAAAyWFUlV9bqqekzNnFNVV1bVMfMeDgAAYLUt90jST3X3HUmOSbJfklcmOXNuUwEAACzIciOppu/HJ/nd7r5mWAMAAFgzlhtJV1TVJZlF0sVV9egk357fWAAAAIux1zL3e1WSI5N8qbvvrqrHJjl1fmMBAAAsxnKPJHWSI5L83HT/UUkePpeJAAAAFmi5kfS2JM9MctJ0/84kb53LRAAAAAu03NPtju7up1XVXyRJd3+tqh46x7kAAAAWYrlHkr5VVXtmdtpdqmpDXLgBAABYg5YbSWcl+WCSx1XVG5P8cZI3zW0qAACABVnW6Xbd/e6quiLJczL7/Ugndvfn5zoZAADAAiwrkqpq/yS3JnnvsPaQ7v7WvAYDAABYhOWebndlkm1J/irJtdPt66rqyqp6+ryGAwAAWG3LjaRLkxzf3Qd092OTHJfko0l+NrPLgwMAAKwJy42kZ3T3xUt3uvuSJM/s7k8nedhcJgMAAFiA5f6epJur6leSnD/df2mSW6bLgrsUOAAAsGYs90jSy5MckuRD09f3TWt7JnnJfEYDAABYfcu9BPhfJ3ntTjZvXblxAAAAFmu5lwDfkOSXkzwpycOX1rv7X8xpLgAAgIVY7ul2707yhSSHJfm1JNcluXxOMwEAACzMciPpsd19TpJvdfenuvunkjiKBAAArDnLvbrdt6bvN1fVCUm+kmT/+YwEAACwOMuNpN+oqn2S/GKS30rymCT/Zm5TAQAALMhyr2730enm7UmePb9xAAAAFmu5V7c7LLNLgG8cH9PdL5zPWAAAAIux3NPtPpTknCQfSfLt+Y0DAACwWMuNpL/r7rPmOgkAAMAuYLmR9JaqOiPJJUnuWVrs7ivnMhUAAMCCLDeSnpLklZn9bqSl0+06flcSAACwxiw3kn4iyfd39zfnOQwAAMCi7bHM/T6bZN95DgIAALArWO6RpH2TfKGqLs8//EySS4ADAABrynIj6Yy5TgEAALCLWFYkdfen5j0IAADAruB+I6mq7szsKnbfsSlJd/dj5jIVAADAgtxvJHX3o1drEAAAgF3Bcq9uBwAAsC6IJAAAgIFIAgAAGIgkAACAgUgCAAAYiCQAAICBSAIAABiIJAAAgIFIAgAAGMwtkqrq3Kq6tao+O6ztX1WXVtW10/f9pvWqqrOqamtVXV1VTxsec8q0/7VVdcq85gUAAEjmeyTpnUmO3W7t9CSXdffhSS6b7ifJcUkOn75OS/L2ZBZVSc5IcnSSo5KcsRRWAAAA8zC3SOruP0xy23bLm5OcN90+L8mJw/q7eubTSfatqoOSPD/Jpd19W3d/Lcml+c7wAgAAWDGr/ZmkA7v75un2V5McON0+OMkNw343Tms7W/8OVXVaVW2pqi3btm1b2akBAIB1Y2EXbujuTtIr+Hxnd/em7t60YcOGlXpaAABgnVntSLplOo0u0/dbp/Wbkhw67HfItLazdQAAgLlY7Ui6MMnSFepOSfLhYf3k6Sp3z0hy+3Ra3sVJjqmq/aYLNhwzrQEAAMzFXvN64qp6b5JnJTmgqm7M7Cp1Zya5oKpeleT6JC+Zdr8oyfFJtia5O8mpSdLdt1XVrye5fNrvDd29/cUgAAAAVszcIqm7T9rJpufsYN9O8uqdPM+5Sc5dwdEAAAB2amEXbgAAANgViSQAAICBSAIAABiIJAAAgIFIAgAAGIgkAACAgUgCAAAYiCQAAICBSAIAABiIJAAAgIFIAgAAGIgkAACAgUgCAAAYiCQAAICBSAIAABiIJAAAgIFIAgAAGIgkAACAgUgCAAAYiCQAAICBSAIAABiIJAAAgIFIAgAAGIgkAACAgUgCAAAYiCQAAICBSAIAABiIJAAAgIFIAgAAGIgkAACAgUgCAAAYiCQAAICBSAIAABiIJAAAgIFIAgAAGIgkAACAgUgCAAAYiCQAAICBSAIAABiIJAAAgIFIAgAAGIgkAACAgUgCAAAYiCQAAICBSAIAABiIJAAAgIFIAgAAGIgkAACAgUgCAAAYiCQAAICBSAIAABiIJAAAgIFIAgAAGIgkAACAgUgCAAAYiCQAAICBSAIAABiIJAAAgIFIAgAAGIgkAACAgUgCAAAYiCQAAICBSAIAABiIJAAAgIFIAgAAGIgkAACAgUgCAAAYiCQAAICBSAIAABiIJAAAgIFIAgAAGIgkAACAgUgCAAAYiCQAAICBSAIAABiIJAAAgIFIAgAAGIgkAACAgUgCAAAYiCQAAICBSAIAABiIJAAAgIFIAgAAGCwkkqrquqr6TFVdVVVbprX9q+rSqrp2+r7ftF5VdVZVba2qq6vqaYuYGQAAWB8WeSTp2d19ZHdvmu6fnuSy7j48yWXT/SQ5Lsnh09dpSd6+6pMCAADrxq50ut3mJOdNt89LcuKw/q6e+XSSfavqoEUMCAAArH2LiqROcklVXVFVp01rB3b3zdPtryY5cLp9cJIbhsfeOK39A1V1WlVtqaot27Ztm9fcAADAGrfXgl73n3b3TVX1uCSXVtUXxo3d3VXVD+YJu/vsJGcnyaZNmx7UYwEAAJYs5EhSd980fb81yQeTHJXklqXT6Kbvt06735Tk0OHhh0xrAAAAK27VI6mqHlVVj166neSYJJ9NcmGSU6bdTkny4en2hUlOnq5y94wktw+n5QEAAKyoRZxud2CSD1bV0uu/p7s/XlWXJ7mgql6V5PokL5n2vyjJ8Um2Jrk7yamrPzIAALBerHokdfeXkjx1B+t/k+Q5O1jvJK9ehdEAAAB2qUuAAwAALJxIAgAAGIgkAACAgUgCAAAYiCQAAIDBIi4BDqyijad/bNEj7JauO/OERY8AACyII0kAAAADkQQAADAQSQAAAAORBAAAMBBJAAAAA5EEAAAwEEkAAAADkQQAADAQSQAAAAORBAAAMBBJAAAAA5EEAAAwEEkAAAADkQQAADAQSQAAAAORBAAAMBBJAAAAA5EEAAAwEEkAAAADkQQAADAQSQAAAAORBAAAMBBJAAAAA5EEAAAwEEkAAAADkQQAADAQSQAAAAORBAAAMBBJAAAAA5EEAAAwEEkAAAADkQQAADAQSQAAAAORBAAAMBBJAAAAA5EEAAAwEEkAAAADkQQAADAQSQAAAAORBAAAMBBJAAAAA5EEAAAwEEkAAAADkQQAADAQSQAAAAORBAAAMBBJAAAAg70WPQAAa8fG0z+26BF2S9edecKiRwBg4EgSAADAQCQBAAAMRBIAAMBAJAEAAAxEEgAAwEAkAQAADEQSAADAQCQBAAAMRBIAAMBAJAEAAAxEEgAAwEAkAQAADEQSAADAQCQBAAAMRBIAAMBAJAEAAAxEEgAAwEAkAQAADEQSAADAQCQBAAAM9lr0AAAA342Np39s0SPslq4784RFjwC7PEeSAAAABo4kAQDAA3Dk8ru3Ox69dCQJAABgIJIAAAAGIgkAAGAgkgAAAAYiCQAAYCCSAAAABrtNJFXVsVX1l1W1tapOX/Q8AADA2rRbRFJV7ZnkrUmOS3JEkpOq6ojFTgUAAKxFu0UkJTkqydbu/lJ3fzPJ+Uk2L3gmAABgDaruXvQMD6iqXpzk2O7+V9P9VyY5urtfM+xzWpLTprv/KMlfrvqga8MBSf560UOwbni/sZq831hN3m+sNu+5784TunvD9ot7LWKSeejus5Ocveg5dndVtaW7Ny16DtYH7zdWk/cbq8n7jdXmPbeydpfT7W5Kcuhw/5BpDQAAYEXtLpF0eZLDq+qwqnpokpcluXDBMwEAAGvQbnG6XXffW1WvSXJxkj2TnNvd1yx4rLXKKYusJu83VpP3G6vJ+43V5j23gnaLCzcAAACslt3ldDsAAIBVIZIAAAAGIgkAAGCwW1y4gfmoqicm2Zzk4GnppiQXdvfnFzcVa9n0njs4yZ91913D+rHd/fHFTQbwvamqo5J0d19eVUckOTbJF7r7ogWPxjpQVe/q7pMXPcda4sIN61RV/UqSk5Kcn+TGafmQzC6vfn53n7mo2Vibqurnkrw6yeeTHJnkdd394Wnbld39tEXOx/pRVad29zsWPQdrR1WdkeS4zP7y+dIkRyf5RJLnJbm4u9+4wPFYY6pq+1+DU0meneQPkqS7X7jqQ61BImmdqqq/SvKk7v7WdusPTXJNdx++mMlYq6rqM0me2d13VdXGJB9I8rvd/Zaq+ovu/uGFDsi6UVX/t7u/b9FzsHZM/347MsnDknw1ySHdfUdVPSKzI+f/eKEDsqZU1ZVJPpfkvyfpzCLpvZn9RXe6+1OLm27tcLrd+vXtJI9Pcv126wdN22Cl7bF0il13X1dVz0rygap6Qmb/gocVU1VX72xTkgNXcxbWhXu7+74kd1fVF7v7jiTp7m9Ulf+mstI2JXldkl9N8m+7+6qq+oY4Wlkiaf36+SSXVdW1SW6Y1r4vyQ8mec3CpmItu6Wqjuzuq5JkOqL040nOTfKUxY7GGnRgkucn+dp265Xkf6/+OKxx36yqR3b33UmevrRYVfvEXzyywrr720neXFXvn77fEv9Pv+L8ga5T3f3xqvqhJEflH1644fLpb8NgpZ2c5N5xobvvTXJyVf23xYzEGvbRJHsvRfmoqj65+uOwxv3z7r4n+fv/gV3ykCSnLGYk1rruvjHJT1TVCUnuWPQ8a43PJAEAAAz8niQAAICBSAIAABiIJAB2S1V11wPv9V0/9+ur6pfm9fwA7NpEEgAAwEAkAbBmVNUPVNXHq+qKqvqjqnpiVe1TVddX1R7TPo+qqhuq6iE72n/RPwMAiyeSAFhLzk7y2u5+epJfSvK27r49yVVJfmza58eTXNzd39rR/guYGYBdjN+TBMCaUFV7J/nRJO+vqqXlh03f35fkpUk+keRlSd72APsDsI6JJADWij2SfL27j9zBtguTvKmq9k/y9CR/kORR97M/AOuY0+0AWBO6+44kX66qn0iSmnnqtO2uJJcneUuSj3b3ffe3PwDrm0gCYHf1yKq6cfj6hSSvSPKqqvo/Sa5JsnnY/31JfnL6vuT+9gdgnaruXvQMAAAAuwxHkgAAAAYiCQAAYCCSAAAABiIJAABgIJIAAAAGIgkAAGAgkgAAAAYiCQAAYPD/AFgh0gALOHF2AAAAAElFTkSuQmCC\n","text/plain":["<Figure size 1008x576 with 1 Axes>"]},"metadata":{"tags":[],"needs_background":"light"}}]},{"cell_type":"code","metadata":{"id":"l-1HKjGKKdwq"},"source":["def discrete_wave_transform(image):\n","    coeffs = pywt.dwt2(image, 'haar')\n","    cA, (cH, cV, cD) = coeffs\n","    return cA\n","\n","def gaussian_matched_filter_kernel(L, sigma, t = 3, mf = True):\n","    '''\n","    K =  1/(sqrt(2 * pi) * sigma ) * exp(-x^2/2sigma^2), |y| <= L/2, |x| < s * t\n","    '''\n","    dim_y = int(L)\n","    dim_x = 2 * int(t * sigma)\n","    arr = np.zeros((dim_y, dim_x), 'f')\n","    \n","    ctr_x = dim_x / 2 \n","    ctr_y = int(dim_y / 2.)\n","\n","    # an un-natural way to set elements of the array\n","    # to their x coordinate. \n","    # x's are actually columns, so the first dimension of the iterator is used\n","    it = np.nditer(arr, flags=['multi_index'])\n","    while not it.finished:\n","        arr[it.multi_index] = it.multi_index[1] - ctr_x\n","        it.iternext()\n","\n","    two_sigma_sq = 2 * sigma * sigma\n","    sqrt_w_pi_sigma = 1. / (sqrt(2 * pi) * sigma)\n","    if not mf:\n","        sqrt_w_pi_sigma = sqrt_w_pi_sigma / sigma ** 2\n","\n","    def k_fun(x):\n","        return sqrt_w_pi_sigma * exp(-x * x / two_sigma_sq)\n","\n","    def k_fun_derivative(x):\n","        return -x * sqrt_w_pi_sigma * exp(-x * x / two_sigma_sq)\n","\n","    if mf:\n","        kernel = k_fun(arr)\n","        kernel = kernel - kernel.mean()\n","    else:\n","        kernel = k_fun_derivative(arr)\n","\n","    # return the \"convolution\" kernel for filter2D\n","    return cv2.flip(kernel, -1)\n","\n","#Creating a matched filter bank using the kernel generated from the above functions\n","def createMatchedFilterBank(K, n = 12):\n","    rotate = 180 / n\n","    center = (K.shape[1] / 2, K.shape[0] / 2)\n","    cur_rot = 0\n","    kernels = [K]\n","    filters = []\n","    ksize = 31\n","    \n","    for i in range(1, n):\n","        cur_rot += rotate\n","        r_mat = cv2.getRotationMatrix2D(center, cur_rot, 1)\n","        k = cv2.warpAffine(K, r_mat, (K.shape[1], K.shape[0]))\n","        kernels.append(k)\n","\n","    for theta in np.arange(0, np.pi, np.pi / 16):\n","        kern = cv2.getGaborKernel((ksize, ksize), 6, theta,12, 0.37, 0, ktype=cv2.CV_32F)\n","        kern /= 1.5*kern.sum()\n","        filters.append(kern)\n","\n","    return kernels,filters\n","\n","#Given a filter bank, apply them and record maximum response\n","def applyFilters(im, kernels):\n","\n","    images = np.array([cv2.filter2D(im, -1, k) for k in kernels])\n","    return np.max(images, 0)\n","\n","def kmeans(image):\n","    img = image.reshape((576, 750))\n","    Z = img.reshape((-1,3))\n","\n","    # convert to np.float32\n","    Z = np.float32(Z)\n","\n","    k=cv2.KMEANS_PP_CENTERS\n","\n","\n","    # define criteria, number of clusters(K) and apply kmeans()\n","    criteria = (cv2.TERM_CRITERIA_EPS + cv2.TERM_CRITERIA_MAX_ITER, 10, 1.0)\n","    K = 2\n","    ret,label,center=cv2.kmeans(Z,K,None,criteria,10,k)\n","\n","    # Now convert back into uint8, and make original image\n","    center = np.uint8(center)\n","    res = center[label.flatten()]\n","    res2 = res.reshape((img.shape))\n","    return np.array(res2).flatten()\n","\n","immatrix=[]\n","\n","for index, row in imagecsv.iterrows():\n","  \n","  img_path = os.path.join(img_pt,row['image']+'.png')\n","\n","  if not os.path.isfile(img_path):\n","    continue\n","  img = cv2.imread(img_path)\n","  img_gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)\n","\n","  equ = cv2.equalizeHist(img_gray)\n","\n","  dwt_img = discrete_wave_transform(equ)\n","\n","  kernel = gaussian_matched_filter_kernel(20, 5)\n","  kernels,filters = createMatchedFilterBank(kernel, 4)\n","\n","  equ3 = applyFilters(dwt_img,kernels)\n","  equ4 = applyFilters(equ3,filters)\n","  kmeans_img = kmeans(equ4)\n","  immatrix.append(kmeans_img)\n","\n","imagecsv['data'] = immatrix"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"id":"869WtG-KQP5P","colab":{"base_uri":"https://localhost:8080/"},"executionInfo":{"status":"ok","timestamp":1606270725494,"user_tz":360,"elapsed":6176,"user":{"displayName":"Gerry","photoUrl":"https://lh4.googleusercontent.com/-oUPavxcTH4Q/AAAAAAAAAAI/AAAAAAAAAG4/Edgkb0AtMhI/s64/photo.jpg","userId":"11854613171556017399"}},"outputId":"712ec190-e93d-44d0-8a33-38b595a97a4e"},"source":["#imagecsv.to_pickle(r'/content/drive/MyDrive/Diabetic Retinopathy Detection/diabetic_retinopathy_df.pkl')\n","imagecsv = pd.read_pickle(r'/content/drive/MyDrive/Diabetic Retinopathy Detection/diabetic_retinopathy_df.pkl')\n","imagecsv.head"],"execution_count":20,"outputs":[{"output_type":"execute_result","data":{"text/plain":["<bound method NDFrame.head of            image  level                                               data\n","0        10_left      0  [211, 203, 212, 211, 203, 212, 211, 203, 212, ...\n","1       10_right      0  [47, 41, 47, 47, 41, 47, 47, 41, 47, 47, 41, 4...\n","2        13_left      0  [10, 0, 10, 10, 0, 10, 10, 0, 10, 10, 0, 10, 1...\n","3       13_right      0  [212, 201, 212, 212, 201, 212, 212, 201, 212, ...\n","4        15_left      1  [56, 50, 56, 56, 50, 56, 56, 50, 56, 56, 50, 5...\n","...          ...    ...                                                ...\n","2995  3706_right      0  [96, 82, 98, 96, 82, 98, 96, 82, 98, 96, 82, 9...\n","2996   3708_left      0  [143, 132, 142, 143, 132, 142, 143, 132, 142, ...\n","2997  3708_right      0  [214, 202, 214, 214, 202, 214, 214, 202, 214, ...\n","2998   3710_left      0  [96, 88, 97, 96, 88, 97, 96, 88, 97, 96, 88, 9...\n","2999  3710_right      0  [175, 162, 175, 175, 162, 175, 175, 162, 175, ...\n","\n","[3000 rows x 3 columns]>"]},"metadata":{"tags":[]},"execution_count":20}]},{"cell_type":"code","metadata":{"colab":{"base_uri":"https://localhost:8080/","height":303},"id":"0AU-L-60xEuE","executionInfo":{"status":"ok","timestamp":1606270739419,"user_tz":360,"elapsed":1136,"user":{"displayName":"Gerry","photoUrl":"https://lh4.googleusercontent.com/-oUPavxcTH4Q/AAAAAAAAAAI/AAAAAAAAAG4/Edgkb0AtMhI/s64/photo.jpg","userId":"11854613171556017399"}},"outputId":"7abdd86a-de97-41b9-bb85-ebf5b00cbb12"},"source":["i = 167\n","print(imagecsv['image'][i])\n","print(imagecsv['level'][i])\n","plt.imshow(imagecsv['data'][i].reshape((576, 750)),cmap='gray')\n","plt.show()"],"execution_count":23,"outputs":[{"output_type":"stream","text":["197_right\n","0\n"],"name":"stdout"},{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"tags":[],"needs_background":"light"}}]},{"cell_type":"markdown","metadata":{"id":"KTrCveUixEuH"},"source":["# Model training"]},{"cell_type":"code","metadata":{"id":"2XOBIbXbxEuH","executionInfo":{"status":"ok","timestamp":1606270759978,"user_tz":360,"elapsed":723,"user":{"displayName":"Gerry","photoUrl":"https://lh4.googleusercontent.com/-oUPavxcTH4Q/AAAAAAAAAAI/AAAAAAAAAG4/Edgkb0AtMhI/s64/photo.jpg","userId":"11854613171556017399"}}},"source":["clf = SVC()"],"execution_count":25,"outputs":[]},{"cell_type":"code","metadata":{"id":"YTyan6JMxEuH","executionInfo":{"status":"ok","timestamp":1606270758704,"user_tz":360,"elapsed":2930,"user":{"displayName":"Gerry","photoUrl":"https://lh4.googleusercontent.com/-oUPavxcTH4Q/AAAAAAAAAAI/AAAAAAAAAG4/Edgkb0AtMhI/s64/photo.jpg","userId":"11854613171556017399"}}},"source":["X = np.row_stack(imagecsv['data'])\n","y = np.array(imagecsv['level'])\n","\n","X_train, X_test, y_train, y_test = train_test_split(\n","    X,\n","    y,\n","    test_size=0.2,\n","    shuffle=True,\n","    random_state=42,\n",")"],"execution_count":24,"outputs":[]},{"cell_type":"code","metadata":{"colab":{"base_uri":"https://localhost:8080/","height":308},"id":"knNM0kLY_G_9","executionInfo":{"status":"ok","timestamp":1606270761132,"user_tz":360,"elapsed":620,"user":{"displayName":"Gerry","photoUrl":"https://lh4.googleusercontent.com/-oUPavxcTH4Q/AAAAAAAAAAI/AAAAAAAAAG4/Edgkb0AtMhI/s64/photo.jpg","userId":"11854613171556017399"}},"outputId":"f404254a-dc87-4c53-8999-50ee1be42a45"},"source":["def plot_bar(y, loc='left', relative=True):\n","    width = 0.35\n","    if loc == 'left':\n","        n = -0.5\n","    elif loc == 'right':\n","        n = 0.5\n"," \n","    # calculate counts per type and sort, to ensure their order\n","    unique, counts = np.unique(y, return_counts=True)\n","    sorted_index = np.argsort(unique)\n","    unique = unique[sorted_index]\n"," \n","    if relative:\n","        # plot as a percentage\n","        counts = 100*counts[sorted_index]/len(y)\n","        ylabel_text = '% count'\n","    else:\n","        # plot counts\n","        counts = counts[sorted_index]\n","        ylabel_text = 'count'\n"," \n","    xtemp = np.arange(len(unique))\n"," \n","    plt.bar(xtemp + n*width, counts, align='center', alpha=.7, width=width)\n","    plt.xticks(xtemp, unique)\n","    plt.xlabel('Level')\n","    plt.ylabel(ylabel_text)\n"," \n","plt.suptitle('relative amount of photos per type')\n","plot_bar(y_train, loc='left')\n","plot_bar(y_test, loc='right')\n","plt.legend([\n","    'train ({0} photos)'.format(len(y_train)),\n","    'test ({0} photos)'.format(len(y_test))\n","]);"],"execution_count":26,"outputs":[{"output_type":"display_data","data":{"image/png":"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\n","text/plain":["<Figure size 432x288 with 1 Axes>"]},"metadata":{"tags":[],"needs_background":"light"}}]},{"cell_type":"code","metadata":{"id":"6PVj5iU9ve5w"},"source":["#clf.fit(X_train, y_train)\n","\n","#dump(clf, r'/content/drive/MyDrive/Diabetic Retinopathy Detection/diabetic_retinopathy_trained_model.joblib')\n","clf = load(r'/content/drive/MyDrive/Diabetic Retinopathy Detection/diabetic_retinopathy_trained_model.joblib')"],"execution_count":null,"outputs":[]},{"cell_type":"code","metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"2LBmx6ebFij2","executionInfo":{"status":"ok","timestamp":1606252377379,"user_tz":360,"elapsed":1004649,"user":{"displayName":"Gerry","photoUrl":"https://lh4.googleusercontent.com/-oUPavxcTH4Q/AAAAAAAAAAI/AAAAAAAAAG4/Edgkb0AtMhI/s64/photo.jpg","userId":"11854613171556017399"}},"outputId":"fe26aabf-ef0f-43b0-c4bd-a8047930accc"},"source":["y_pred = clf.predict(X_test)\n","accuracy = accuracy_score(y_test, y_pred)\n","print('Model accuracy is: ', accuracy)"],"execution_count":null,"outputs":[{"output_type":"stream","text":["Model accuracy is:  0.7483333333333333\n"],"name":"stdout"}]},{"cell_type":"code","metadata":{"colab":{"base_uri":"https://localhost:8080/"},"id":"LwtqyCOOLXNx","executionInfo":{"status":"ok","timestamp":1606252440974,"user_tz":360,"elapsed":805,"user":{"displayName":"Gerry","photoUrl":"https://lh4.googleusercontent.com/-oUPavxcTH4Q/AAAAAAAAAAI/AAAAAAAAAG4/Edgkb0AtMhI/s64/photo.jpg","userId":"11854613171556017399"}},"outputId":"ebbbd5ad-b579-4494-d69b-0b5b3df21b1b"},"source":["from sklearn.metrics import classification_report\n","\n","cr = classification_report(y_test,y_pred)\n","print(cr)"],"execution_count":null,"outputs":[{"output_type":"stream","text":["              precision    recall  f1-score   support\n","\n","           0       0.75      1.00      0.86       449\n","           1       0.00      0.00      0.00        39\n","           2       0.00      0.00      0.00        87\n","           3       0.00      0.00      0.00        12\n","           4       0.00      0.00      0.00        13\n","\n","    accuracy                           0.75       600\n","   macro avg       0.15      0.20      0.17       600\n","weighted avg       0.56      0.75      0.64       600\n","\n"],"name":"stdout"},{"output_type":"stream","text":["/usr/local/lib/python3.6/dist-packages/sklearn/metrics/_classification.py:1272: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.\n","  _warn_prf(average, modifier, msg_start, len(result))\n"],"name":"stderr"}]}]}