isic-images-analysis-with-dask (1).ipynb

{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"markdown","source":"![image.png](attachment:26316104-3829-4733-9f91-abda2c67ea86.png)![image.png](attachment:bde69095-b9e1-4aaa-b68c-ac49898439da.png)\n\n[https://dask.org/](http://)\n","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19"},"attachments":{"26316104-3829-4733-9f91-abda2c67ea86.png":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAWAAAACPCAYAAADEBAGvAAAgAElEQVR4Ae2dX+gux1nHf5fhkMA5JyfnxEAaSxuNsYXQ5qR/YlpPjdEIx6OpSYykbQ5RS2kTU1oiUjCNpqAXoRZKlKKEpNEWSsEbNXgTyFUR/4SAVUIpabyoRC8KuerVK585v+/r85szMzuz7+67u+/7LCzz/tmdmZ0/n/3uM8/MHlx33XUr370MvA14G/A2sP02cOCFvv1C9zL3Mvc24G2ANuAA9icAfwLyNuBtYKI24ACeqOBdAbkC8jbgbcAB7AB29eNtwNvARG3AATxRwbv6cfXjbcDbgAPYAezqx9uAt4GJ2oADeKKCd/Xj6sfbgLcBB7AD2NWPtwFvAxO1AQfwRAXv6sfVj7cBbwMOYAewqx9vA94GJmoDDuCJCt7Vj6sfbwPeBhzADmBXP94GvA1M1AYcwBMV/BDq59S1P+EdZ8H1N0Qb8DiWraIdwAvtwMD3Ex9618ohvOwO6ADd7/pzAC8UwAfHTq1+8JUHV4Teife7E3v9L7f+HcALBDCq98JtN61+9LWLqy/e+4HVwfHTDuEF1qODc7ngHKruHMAL7LgAF/CigIMKPjjhAF5gPQ7ViT2e5YLcAbzAjovZ4duP/XKA7xtf/fglFXzyjEN4gXXp8FwuPIeoOwfwAjvtwcGJ1StfuncNYCDstuD97shDwMDj2H4bcgAvEMBXnjwd7L+YH4Dv//zVw66CF1iPDrztA29uZe4AXljHBb64n8n+C4DZv//nD61QxnNrYJ4fh4y3gXwbcAAvDMAHJ8+svvyxOy4D8FoFu0eE34QW1qb3GdAO4IU1VjsAJxOEq+C8wtjnzu3XPv924QBeGoAPTqzVrwUwJghXwfPvcA5FryPbBhzACwIwEzDee+MNRwbgrPoFwG4L9g5uO7h/nnd7cAAvCMDYf+0EDMFXg3BSwb97/uyKwTrvfPPufF4/Xj8O4CUB2EzAsOYHC+AfPvfJ1Utfus89IhZUrw7i/QWxA3hBHRU3M7mfpQAMfNn/+4VPrX72nTf4SmkLqluH8H5C2AG8kE6q5SdZgEcQjk0QAjDhs4/d7bPjFlK3Dt/9hC/17gCeaSdlwZ0nLn5kbcvl+/OfvmsN35QCxgYsCPtg3P52agf6cureATxDAB8cXLn6x6cfXJ17z0+tzQiYH77/p/clAQxsLXxlhvjNO29ZA9w75XI6pdfV/tSVA3hmAGaiBcr3e9/4zAoQ0xllfpDpoUv9CsB/8/u/6maImdWvw3V/4FpT1w7gGXVQXMdQrf/7t589an6IvB9iAMv/V+YHAZjBOEG8pjH4MQ4HbwPbbQMO4JkAGPjiuQB82e3COth/MT9YE4QG4IAvS1MSxgB+81uPhjdnuE/wdjuVQ8zLu7YNOIBnAGBMDEDyjW8+EuBrTQdhMO7C2SyAge9nL9yWBXDwhvAFenxSygzaeS2U9uk4B/DEDRP4YvfF5iv1e/311x8ZfHvlyXuOqF+ZIFC9n7vn/WF6ckoBY4L4l6/6MpX71KH9Wpelvh3AEwI4wPfgRICk4BvU7+G6vqhi1v6NzQ8CMGYI4B08JIwLmrUBY4ZwE8SyOqVDdH/qywE8EYBT8AXCdgYbYP27T/9iEsAoXtYFxkQBhJl+LBswylc7APa1IfanQzu8l1XXDuAJABzge/x0UL4Akh34WvXLMbx6Pqd+ATCd7cyZMwHCLNKTA7C1KXsHXVYH9fra7fpyAG8ZwJgDsPEy4Cb4CsBAF6DS6UrqF9MDb0VG+XIs5zFpIwdgtwPvdid2SC+3fh3AWwQwwASUmAdi+P7ZI/+/dgNAtbZf636G/Rf1y7rAHKfOB7BjANs03B94uZ1Udezh7tWhA3hLAAaQ2GKtbVaARA1bQHKsPB9kgrAQtupXnZJzsAMrfgt5zBsxsHWeh7vXqb1Ol1OnDuCRAYxKBa744+bgaNdsYNH1J4zfrwCM8mVnNbQUTFHXpTRwV3NviOV0TIfoftSVA3hEAOOhQEeKlalVpyy6g3pVhwPA3/vjj4Zd8O1Sv5wLXIFsDvLWxKG0PNyPTu71PN96dgCPBGCZHGSXJYzhGEwPhwNpdBJU7PMXP3wZfC2A7SQNDehxrgbi4jQ0wBdAb9LyTjnfTul1sz914wAeGMDBL/f46eClkJudJgVsTQ8AFLezkvqV3y8dFPhiUyYODcbxWw7AfT0h8Mq46qqrRt2vvvrq9RPAHODDNbfsU+aZsqN+rrjiiiO76oz/2U+fPr32sBk7v7VlN2Q+atPUcUOmvUlcDuCBAHzJ1nsirMvA2rwxfK0CRpWGNRqs6eHYqdWrXzhfBLDczqhw7MpMXwa6agCo7jgdKeB4oE/nlEIa69mzZ1cXL14cbb///vtDGgcHByvtwIS0S3kb6z/SBVg8adTuwG6s/MTxkj+VE6Hq5/HHH1/ZXXVG+Z47d2518803HznPxhF/5noogz51wDm15dYn/rg8+E5eyXNtupRFKp4pfnMADwBgwMjAGLZewJtaIF1gBIiX1OiltX6pdGt6SClgBt5wSxNsUdnYe/FuiL0niFsKm7QE4PjYmsYGCOnU29refPPN1csvvxzSFBRQbjV5HeoY0v3Od77TdMmUEWU1VB5y8ahMnnrqqdVrr73WlEd78I9//OMVZf3666+HeCjzF198cfXMM8+siBt4A22lp7DmRsOxNRvpc2zuWmt/p30A3paNshsi7do8lo5zAG8AYECI8uVVQYBXS0SmAGxNA5wjswGDbg/dfvNa+aYAHNzOrFo+OBHge9kA3uGU5ByAaagtqmPbAI47EVAgz9vqLJQNkGndAPbYeSR+4LjtDVgDaiANmLsgTD5rNuIcosyIo/VmxDVs+8aeg7ADuBHAwdRw/HRQo9hk7cI4NQBmIoaULHGx9oOgq9B6P/CZyhM4MTMwtRhFG3s2oKT5Lwdg+4qjXIOwv08NYHVklBkdTWVg8zjkZ9J4+umnlWxTOGb+iLtVlTdlvvLgZ599thOa5LVmGwLAferr0Ucf7byGIdtUV1wO4EoAB/AeOxUUmcAr31yBl1DqlxCzg0wPmCeIQ/ClYgCmtfumAMxav6hkjudcIAp82eNFdogP27JMDwp1vB3062oY/D8XANOh9diIva8m732OoUMDhj4b6pTz+6RbOoc45wBfykQ3wq781pTfpgCmHWD/btlQ8WPUUak8uv5zABcAHKB7kkGPE8FDAVOAdQmrBTD2Wmy1VsGl7L4xgC83PVx5ZOlKu3IaFY1JBFUs8CoUgGNgdzWOOQGYjobdcKzHRzo0ds++2xhmCGAxhdkhVwZzAjBl03KzxJTCObYPdrX/bfzvAI4ALJUKIPnMrLR4WrDAqzBWwNiD2TEHoFoBo61Mvj994dbVfzz+K2EXeBUCedK055EfwRWg/te3fu/IABzxc/yfPPyRLIBbZ8PNDcCC8BgdiThfeOGFHHuqfufmMGQHJ0+AYy7bXABMuWAOadmwX9OebT+cw2cH8OEkBh7zLXRZh1dAJLR2WYFXoQUw4OU1Qfj0opzjDqlBtxx8lZa118r0YBVtPAAnAPNGZR2nUAqY/6wJpKsBzhHAdLoxHiWHgB32Y+LpKtea/4lnTuqXcp8DgLnJ3XXXXS3sDR4eQ9VLTd21HLN3AEbVrk0Lh4uZA0sUaQzdHICtGULwBbzYfYPN9uBEEnQH11y7+o333rhWvkDYpiH4orpj9UvcFqjxAByVzjkxgAVfwl0BML2PwZShFA2dmvg23WSnbumAuWMBxlxsvyqXOQCYcsEUVbu99dZbszQ9qN53CsCCq0KAF3bU7SFssZvi9iXgSokqjIGo71YBpwDMgugo3pzCRPne+dM3HIFvCsBMRSaedQVFq5wBYWCaGlCbGsA09pa9thOljlPHUjltEg4JO9zmhnBxIk+tm3x5UeLagSY7j+zsX//614OpBXMLx2vnqQLgs3MjYcfGyg7wMIVMDWDKhPy2bExE4Qa7SfsY89ydATDgA65ADpX5ife8c/WHv/DuANq/vvd9q5d+5+dX//7YXev9u5+/+zIYpoAoVWoBLNMDpgaB1yrWuMJy8I3TQ4FjBtH5fNabLqxrGQC2EzDWxydswNtSwFJ/dJKWHWDxqN2iatQBh/A8wETUB3bKQxzWQEr1lQsBOLO1ajfKrqXMNzk2l2f9XluWrV4QPO0A05aNmwz5Ud7mGO4EgIHff/72HUf2Vz91bmV3C18+A+AYwlK7cWjhy6w0wMvMNJRqCbxUuMwOSlNKW6HSwh0NUKuRcEPBNIJpQ65sUr+5dR3ISzwIZwE85iCcAKz814bARkAAXi2bwFObVuo40u7r+5vKaytYUnlCsTFoVLthE52LyhsDwLpJtgxIDtE2UnUz9G87A+B/e+j2FfsrD9+x3rsALAgqFAxtaOGL/y9TjlGmOVODrSANuAn+MfBJV2mhBDGdcD4hah77svUlFoCD/TfyrOA88iVPCWsvFoTHdEPrC+Aj5dVj4GnTR0yAATSH3FCvm/grtwx+DmmKsXXR9/MYACZOTCQtGy6Fc7kplcrSAXzoCmZhKCgSYpMNapfXvx9OPS4VqP7j2D/4uZ85YvLIATheYJ1zNb1ZEzpkggCmdjad0iMEwLi+Cb4CtgCcshvb8+PPLSAYAsDhGhrtfNgE6aBx3mu+93Hmr4FAzYyxUv6mKPdSflr+GxrAxNc6QLpp+bdc76bH7hSArfrlc04Bl8wPmAJi6NaoXVsRgPAvf/3sGr45BQzgUzDFtIH6FXytCaK0qhnpMuMuB2Dr2mbzm/s8BQj0uFkDOo7ZRAHSuWt9f3n8pWPXbJs+/k5R7rk20Pr7kACmLVAWLdsQJqDWa97k+N0A8MkzwfxgTRAWvnwWBAllciAUcHH9knkB04HMAS2FyznYfBn0U/pKN6V+GTC0dl/SAva4ssmtLTZBoHABbSpfgNsCWMpXIfmjUafOTf02FQhawEjn5JG/j+cB6dTaFVHaLVNfeQTua4aYqtxTbaD1tyEBTFytrnjUUd9yb73WIY7fKwDjCYEyxTsCTwkKEJgBwb7QVSUAXoBqlbfgSxgDGNinlDUQZZDPKmDMD+yAtGTH5VwG6KSABV6FKc8J5T8VTgWCVi+APt4QrQNd2JoBAuq2ZttkBH6qck+1gdbfhgIw8bQuhcpgKue15nnK43cDwMdPZxUw0MWmut4NbPuo3Liyguo9tPfiiVED4HgNB8VJXHg+4GkRA1hQBbI6Pg4BrI4jFHgJU1OX4/Pj71OCgI5UOzjWxw5M/LXqSiYFzsGPtmbTOXGZ1nyfstxr8lc6hjKq2UqmAm7ADEq3bEONQZSubYz/9gPAxr1ryEJE9dJQ/uG+24ILnIVvyuzBzYD0c+AHrizAA3xTAC6ZH2RayAE457pWKo8pQUBHrnW6b+18rXZmDerwaNuyYA/uYX0eh1vKvQSyUt2O9d8QACYO6rRlox/2MUONVQ618e4UgAGgdtlgscfGdtbawskdF1TvyTPByyF2f1O6MYDDJIuCF0WsfmMAo2JLg2iYM/ByyAE4tXZE7vr0ewsIWiGoNHIhadeuhdCqNungtXEDAWtj5txaM0QfZU55cO0tI//kSTfgXHlu63fyUrPlbhyc3+qX3ccEta3y6EpnNwB8zbVrBTo2gFG9t/zk9au//7WzIc0aADPtOTdwpgpC/QJpZtmlFHCXgsXEUloHIrV2hNLOhVMCGOXYsugKHTd3HfHvHFtr3ohvLJxb6w3R10MD+/QDDzxQw7FwjOzN5G2InfT7Ap30azaVa5zfloFO0sGMRBxxHS/l++IBjHLk8UP21xjADLoNoYABL2n9xbl3r/71tz643rsAzLoTwLHUIFCv+Bpj+xWArQsaqrbLhxfAl3yAS4N3ubxNCWAAAIRrNzphDTSIs6WTx+qq9XxA2johoI8NFNgDI3YmLfTZUezMRkR9C4xcb659pH7nvJqNpwjKFrWrne+1N0alQdku0fSgstsJAON9kANwUJ/XXNvUiFQ4hICX/Y/ef9Pqnz/6vrDXABh7L4NtNfBH/WrN4RjAeD90qd+Qzw4PiNzAn73W+POUAL50TXWdmc5Y634EIGp9f4k3BXZ+qwVFXzMEadS6yAlGQ4fAHJs3eYnbRu47x25r040il5cl/L4TAGbhHQtga4fF5awGgraygo03AV4AbOHLZ+t7jPombZkciMfGm/pM3vBBjldYkwIGwKnJGnFcQDxn/8V+3OqCRvxzAHAthGoHvAAEarFmy8GTOGrXrSD/LQBTvXIOAJzDJhOH8lYKyfc2NhR+n3It5X2K/xYPYNQpU35zAAbONSCk8AVebLwo3n86f2vYpXxjAMv8QKj0U5MrShVLmsyIywGYiRXAtRQHJgzc13IA7jMAR3pzAHAtLGvWhOBxtWWRm5z5gEfeltXKat4mHNcvZd8yEDc29LgZ1ABvGwDWTa3G5BSX69y+7wSAsfPK9isVKhUMEGsADMjPv+Ntq+c+ePMavLUABr5BaRe8HFIVj92WgTct+KNlLq0HRI3pABtzaRW0PgNw5HcOAK71OMjB0pY7cKhd1AXwl4DDf7VmiL5qjTRqr39sABN/jRImz2Nv1DVt09btUj8vH8AnzwQ/XGsKEHwJa+GbAm8XgAEvPsAoZgDe0ghQrQzQaeEfKWB5QDAVmTccd3lPkCYK2U5BRgnbSRhdA3i5fO8SgFFLLXCQ72+ubIir1gwBkDg+F1fudxR76xq4Y8Ova4p1Sxn3zWs8MJorvyX8vnwAH86CyyngLg8EKglIP3TT2y9TviUA44aGeQMbbg3kbWPgeHYL3xjA2IBr4Eu8ABhbcc4E0WXCsHmzn+cA4KFMEICBjlu7Wd9fWyb63GqGYFptH9VGvlun5NZeY5/jutT8NgBMvqmfVg8N1d2cwsUDGJDJ/ioISwFrGnJNgaNgBdw4jG3An7vlxqalKeP0ZXrIARj1y4I8qOT43Pg7x8QTMKwCrvGgiOPU9zkAuBYSNcqsdnaVfFRVDrkQ2NTGWWtDTaVFOjx2196Masus73HkJ2d/3RaAMf+QVqq8lvTbogEMfFlUJwfg4ANcaRoAwPj4luAbwMtaEpVxphpCmDBx4WxYkS0H4JqBN8UNzGP/XwvgvvZf4l8SgEuKFaXUMmiG4qRz1+wtqpr4cuBSfeZCzBGcT3q10O8L2K7zSmqePG5rY10O0suV2RJ+XzSAAWGXB0SNiqSigDkDdjkAY3KoMWeUKl1paDlMC2A7AKf8lOLSf7iXlcwPpenLiiMXTgng1kf8Etz4r3YRHeCB+1ntXjuoR7zAs48ZwtYP16Id1ztgiC0am3WfnYG1Vne30mAceavZGFxkAgZ51873WrdDpVHj/WLLb26flw3gk2dWz//SLUd8cWV+IIzfNNFV+AD9pTtvOQJha35gxh0Q7Yon9T/nsdtlKS2AsQEzEy681r5j5pzi7zI/fO8bn+nl/6v4pwRwq8sYHV/5jkP+m8Pj+yZmiPia+I6yp44E5E3ClieE0nWQh5pNJoQ4z5haWjZ5q/R9skiV6zZ/WzaAC8tQAuDaQSwVOAC2/r9Sw4IwJoi+AEY9Y5MWgAVfQg3AsRJay4CZzA85BVxaPU3XXAqnBDAds9bLADXF8alrAeStnboFAK3Hks85woJ81T4llGzkxFOzCcBxnXE+Crtl6xoYjNOY0/fFAhgQlqYg91kFjTjxBRZ4CQVfQtR2H/svoCQ/9m0cMYBZhL35hnHsVDA/5ABcM4Ou1BinBnDto3EXEFrMBC0dv8+xmCGATKncp/ivpa5z8CTfmwJYcZBGy7bUacmLBbDUqvX/tZMw+kxBDpUfeUNYADP1uHVaM8qXwUC9HSNl/0UBN5tLjp9e8Zp54JsC8KbmB8qipVOWIEhcrXttR6aT5qYMozRb4mnp8H2PHbqcWss1dzx1XTvzbmwAY1rB3Ne6cW1LW5hnuQBOTMCwAG4FmhomYLeTMmIA186sIz7gy40Ac0j8WiIpYOAb3rrcuGg8g2+4mOUAHLwfKm3JuvY4nArALenSSXOqEvjy39y2OS4ePicAh77T423Ic725xf3Kfl8kgDEVaAnKnALu67EAgLH1xuYHQIwCxuuixgxB+vLQEIBTJogvf+yOZtODBt8EX6uANQOOSt7U1tgCwiEbP+BsmYKbWwmNeMjX3DZG+8mb7YhTf54bgCkPyoinm5aNcYO5lW2pbhcJYAAouFkAAzr2Fv/fuHCAO1OLcwDGV7gLwBa+UuUpBRzektGx0E6cv0sN89LUYwE4ngHXd/GdOK0pAEznAVC1mxZmifPe6sZWm94Qx5Ue4ePr2Nb3OQKYa6c9tNyMqR8m5TD4uq2y2ySdZQL4+OnL1n8Q6AAwkzMAad+CAbAaiJMJQstQBn/ggrkA+OJJockhuilYAGN+CPDNvF6+lG+ui8E1wdeqX03A6Lv2Q5zutgFMZ2tZrYzOlhsBJ67aN1cQDyDfdG+Bc2niSFwP2/jeUtelGwjlXrOV4rDXC0gBastW8oqxcc/h8+IALIUK4Kz6tQDua35QhQBg2YFjAJcG4kgX9W3zlgLwq18433sqsxbeyQH40tTjK3vffFQGhC2dUiYIzB61OyqVDkanZa91O7OdMbfUI/G1+P4qD5uELfbmrsV+bD1s4zPXXet1UoIn8dRspTji6yXOlqci0s8NzMZxT/19cQAGjlKYKQAH80NBodYUuNKwZggpYOAaD8RxU+AcvR2ZfClvFsB4QABfjmevyYs9hnOkfn/43CfXKlgmCOy/eEbUzv6zcac+9wEwI9g1O/DFdgtA8T9tgaU6eM78ANRbVhEborO2elzUqjTKkuvRbsuWMmTXDS9VhzW/AbiWJw/dbFNxjwFg0iHe2huE2gfXRBtO5XMuvy0PwCfPBLhZyKF+2YHjpuYHKgbQsTqa1K8G4IAwadhF3jkWm3EMXwtgzA8yQXA8e58GIPVr4WtNEG9885GNZr7FeWoBsBr9NsPcgAudtWXwhim9AC6+/tbvrZDIDR4qXWBL3nAPAyZMKOE7O4/lnM9gNMdTV6TfZ29R7tRv6YZF+jVbiwLm+nSD46Zbu+kGzbkq07mFiwIw4NLrh2IAS2luan6ggkgHlZsDMAocxcuxKeVr80a+gC+z4DaBrzwfgG8KwEOrX65t7gCms6c6Vy0E6Mi1SrSm41Jetb60pF1aUyG0rUY7dg2YYjt3zTnxMbkbn/IcH5/63gpgtceWJxvSLU2brqnTsY9ZFICBK0pTgJPKlP03TL7YYKUyW9iANQVgVLCdEcdxmD1snvRZ+QrLYvZYN/hIfg5OrPSeuBSAh1a/pD1nAGvFMltGffJcgkkcd833IeFPXH3s4ingDfkbChx1niqP2uvvA2DSI/6WwVWuO+cnnsr/tn9bDIClSjXAZSEn0FF4HDdEIZYAzE1AM+I4zrrEKV+E5CtMiW58VVGcf6Yo88oh1gmOFbA8H/q8dj5OJ/4+VwCXVA0dtMX3l2tMqei4LGq/k36r+aMEs7kBWIvf5MpjbACTLmkA8JYNr5NcOeeuZRu/LwbAJfW7Bt2Gg2+2wAErLmfW/quBOAtggB+bRYAvN4owILjhbDTi571wFr5WAQPgIT0fbBnMEcAyGaSgyYBUyxTWEshtObR8bh0ALJkhAM3cANz1xLANALfWM6Duq7hb6r7PsYsAcJf6BXaxZ0KfwrDnAGBMDSkAA1jZmlMAJj9hxtyG8CU/duBN6jcG8CZr/tprjj/PDcDqRHTAOK+Xyqrt8ZSBrSEG3+K8ACFuFDVbSVHODcClvKoMtgFg1XWLvZ26mOMC7osAcEr96hEf9RsU6QCwUyMKFXz4howYwFK3AJrjUjcHbgb638bZ+hnTwxfv/UBQv9b+KwCjfsOSkz1m09XkZU4AllLMwZfrGQp8NWVTOoZ81C7tCBhyi4rPDcBdr31SHdTceHQzLZVj13+UT4u5p1TWXWmN9f/sAQzg9Noh4Gd34DuG+qWwASjeDjkA86hL3thxQyMf3Aj026YVRjwsKITpIQVfTcRoXcKyJV9zADBmAjo+na2U99ZH/zEnQmBrxEWsdsu5dc0FwHhOMPDWVQeh34zkhpare/JU+7RBfUjFp0xYuTTG/H32AM6pXyngodQvwLMTGLoAbFdbw0YbXNM2HGyzFY3pgTWCSwAeY+DN5mHbAKajo4wAEiPXQJUOVjN4wnEt6/6OPRWY/HAttRvH27LXZ35HIWN7pVwYYAQi29p0c6g11ZDfmk22fF1n37D1ZkfexrD9983/rAEMBHMeBjIFbGr7BbxAnvCJC2fXECZtVkXLKeA4XY7vWwnxecCXt2MIvikF3PLizjj+lu8oBXYe/cfY6UCAno4b760qhbjiOHLfa6DeUk6pY7kuwNW1l8wqxKsbUXwtlA/KFFs2bnlM1wXUmGsAJzckYAO02bkhAD52IM5u/YL5znGcKzc/0mypB8pVE0V4cintXdedKtPUb5QPTxyltOx/5G+otFP5aflttgAGiFyINTnEn+Xi1XLBHBugy9uNj50KU3ufv/jhFQvkvPLkPeulIVsB3JqH3PHW7vvGVz8ezA8pAKPWWzpGLr2a30ln7L0mHzXH1OazJq4hjhkzP8QN8ABQ7iYWQ7v2O/ERf58yIE9d+9AAJL6uNO3/fa5rjHNmC2BUqV64GYNX6hfbq0DdVTgWuhyL2mVdBi2MPgcAc80XbrspvJwT+ArAsffDUKuddZWZ/9/+Jg8vMy+zljYwSwCjPlO+tQIxA141i+5Y6DJoBnRZBtJC137uUsA2/dgE0VLoqWNRtNiVBV6Fsfp99rG7m17cmUrLf3NIeBuYRxuYHYCBJrtglwtjm2uA7TXXhhlqKElmquE98fSFW8M6DHoXG6GFrv08FYDJOx3iB195MOyCb6yAL024ONHrsdA73Dw6nNeD14NtA7MDMPDEsyEHXtQv6liQJQRgKFLWgkAZs8JTVmkAAAXKSURBVPaCXYHMwjcHYN7NFgOYAUA7CKc8Den6Rt5Rv3g8lACMEuZYwdpWon/2Tu1tYJltYFYARtWWvB4EQI4BtIBaq6ApFHgVfvfzd69qAIwStm+pIC+4lmn6MaHSB8At9udc5wCmKPUSfGWC2PQV87k8+O/L7Lheb7tRb7MBMDBCxQI3ga4UMgmDXeBVKPASAt9aAKOA8YZAUdO4ATDvf8sBeFMlyvlB+T55z1r5ooBj8wOuaPj7Kl/e8Xaj43k9ej0GzsyhIQAj9hJw4/9qARyr39gEgScEbyZmAMzOKgPAeGHEANbsO62G1qf8BG9MHoBfpocYwMCXqcg2X33S83O8s3sbmGcbmIUCRt2V7L45+NYo4BjAmBqALmqXQTrgBkwFRTVUAMxqaDGAyQvp9gUw5wF7wKs9BWDg6x4P8+w0aiMeev1s2gYmBzDwxZ5ba3oQAFsVMANzeESsodsxbTgGsG4CpBumP/dY+pJrxc9X4FUYAxj4Pv9p5t67x8OmDdzPd0jOuQ1MCmAgV/L3FfTiUPAllO2X0Np/AS5gf+j2my/ZdDNKN1c5KNWc+g0z8BoBjNLGDxnoosIFX0ILYODLNGSHr4Mj1zb9991pG5MBmEd+DbqhfvvsABjQAkTULd4RxInSDPuhaSE2L3Q1YI5n4kYKwNwMADs3j654+J+4gK+mO8fwtQD+0dcuuvK9bnc6V0378GP2u74nBTDqF2jW7BzLDmBxAUOhWtACRLwKWmGb6gDEcf4db0sCmBtF7bvnZO+Npzxb9SsAO3z3uyOm2qH/tvttYjIA07gAXVCIzGDTLDZ9jkIdq3DMxkleuCmkFDAA5kZAPnJ5CNd0aHJA8cZ7DGDgG7wd3OabLdNcWfvvuw+pXa7jSQE814IFwCUf4NIkDFQv6wNjGold3gRiC2Dsv5/40Lvc1cxND37z2cM24ABOVDoALrqgJV5/JNWLecK6vgm6NrQAZoYbppS53ow8X64wvQ2M1wYcwBkAp9aAYAAudkETeHFvk+qtATCTMGjY2K29gY/XwL1svWzn3AYcwBGAASoDfSn7r/WACOA9fmkJSbwwNPXZwjdlgkD9MvMONzPimHPj8Lw5vLwNjNsGHMARgDE/lBbhYQAOtzLswAKv4Fuz7oTbe8dt0A4ML98ltQEHcALAOfsvHhCoY8CrCSCa/NG18A8rraF43eTggFgSIDyv47ZXB3ACwNb+iylCM/E0A0/wjWffxQpYA2/MxkM1u8lh3MbssPDyXVobcAAbAANITAx9ARzbf5n9huLFNW1pDcPz6zDzNjB+G3AAGwDL/zcFYKlfu/6EzA+yAQvAeEOEKdHHTjl4Tfl6hx6/Q3sZL6uMHcAGECjVGL4yQXQBGPhignBzw7I6gAPL62vKNuAAPgQw5gfWf4gBLPsvoVW/sf03rA9x7JQPspkb2pQN29N2sC6hDTiAD4FRMj8IwjGAMTXILc3tvN7hl9DhPY/zaqcOYANg1G9OAcsEgSsabmha2N3dyubVoB0wXh9LagMO4MNV2R666e0BvikAA10AjJmBhXZYu8FdyryjL6mje17n2V4dwIdvQOYFnDF8GYBjVbTg0cAi79dc6+B1G697tngbGKwNOIAPFbA1PwBdBuTCgu8O3cEam6uweaowr5fp6mXvAYwpAfPDWunypg2HrkPXVZ63gS20gb0HMHd/IOzQnU4FuALzst/XNuAA3sJdbl8bl1+3g9XbQLkNOIAdwP6o6W3A28BEbcABPFHBuzIoKwMvHy+ffWgDDmAHsKsfbwPeBiZqAw7giQp+H+7ufo2uYr0NlNuAA9gB7OrH24C3gYnagAN4ooJ3ZVBWBl4+Xj770AYcwA5gVz/eBrwNTNQGHMATFfw+3N39Gl3FehsotwEHsAPY1Y+3AW8DE7UBB/BEBe/KoKwMvHy8fPahDTiAHcCufrwNeBuYqA04gCcq+H24u/s1uor1NlBuA/8HnH+QJpRUfv8AAAAASUVORK5CYII="},"bde69095-b9e1-4aaa-b68c-ac49898439da.png":{"image/png":"iVBORw0KGgoAAAANSUhEUgAAAWAAAACPCAYAAADEBAGvAAAgAElEQVR4Ae2dX+gux1nHf5fhkMA5JyfnxEAaSxuNsYXQ5qR/YlpPjdEIx6OpSYykbQ5RS2kTU1oiUjCNpqAXoRZKlKKEpNEWSsEbNXgTyFUR/4SAVUIpabyoRC8KuerVK585v+/r85szMzuz7+67u+/7LCzz/tmdmZ0/n/3uM8/MHlx33XUr370MvA14G/A2sP02cOCFvv1C9zL3Mvc24G2ANuAA9icAfwLyNuBtYKI24ACeqOBdAbkC8jbgbcAB7AB29eNtwNvARG3AATxRwbv6cfXjbcDbgAPYAezqx9uAt4GJ2oADeKKCd/Xj6sfbgLcBB7AD2NWPtwFvAxO1AQfwRAXv6sfVj7cBbwMOYAewqx9vA94GJmoDDuCJCt7Vj6sfbwPeBhzADmBXP94GvA1M1AYcwBMV/BDq59S1P+EdZ8H1N0Qb8DiWraIdwAvtwMD3Ex9618ohvOwO6ADd7/pzAC8UwAfHTq1+8JUHV4Teife7E3v9L7f+HcALBDCq98JtN61+9LWLqy/e+4HVwfHTDuEF1qODc7ngHKruHMAL7LgAF/CigIMKPjjhAF5gPQ7ViT2e5YLcAbzAjovZ4duP/XKA7xtf/fglFXzyjEN4gXXp8FwuPIeoOwfwAjvtwcGJ1StfuncNYCDstuD97shDwMDj2H4bcgAvEMBXnjwd7L+YH4Dv//zVw66CF1iPDrztA29uZe4AXljHBb64n8n+C4DZv//nD61QxnNrYJ4fh4y3gXwbcAAvDMAHJ8+svvyxOy4D8FoFu0eE34QW1qb3GdAO4IU1VjsAJxOEq+C8wtjnzu3XPv924QBeGoAPTqzVrwUwJghXwfPvcA5FryPbBhzACwIwEzDee+MNRwbgrPoFwG4L9g5uO7h/nnd7cAAvCMDYf+0EDMFXg3BSwb97/uyKwTrvfPPufF4/Xj8O4CUB2EzAsOYHC+AfPvfJ1Utfus89IhZUrw7i/QWxA3hBHRU3M7mfpQAMfNn/+4VPrX72nTf4SmkLqluH8H5C2AG8kE6q5SdZgEcQjk0QAjDhs4/d7bPjFlK3Dt/9hC/17gCeaSdlwZ0nLn5kbcvl+/OfvmsN35QCxgYsCPtg3P52agf6cureATxDAB8cXLn6x6cfXJ17z0+tzQiYH77/p/clAQxsLXxlhvjNO29ZA9w75XI6pdfV/tSVA3hmAGaiBcr3e9/4zAoQ0xllfpDpoUv9CsB/8/u/6maImdWvw3V/4FpT1w7gGXVQXMdQrf/7t589an6IvB9iAMv/V+YHAZjBOEG8pjH4MQ4HbwPbbQMO4JkAGPjiuQB82e3COth/MT9YE4QG4IAvS1MSxgB+81uPhjdnuE/wdjuVQ8zLu7YNOIBnAGBMDEDyjW8+EuBrTQdhMO7C2SyAge9nL9yWBXDwhvAFenxSygzaeS2U9uk4B/DEDRP4YvfF5iv1e/311x8ZfHvlyXuOqF+ZIFC9n7vn/WF6ckoBY4L4l6/6MpX71KH9Wpelvh3AEwI4wPfgRICk4BvU7+G6vqhi1v6NzQ8CMGYI4B08JIwLmrUBY4ZwE8SyOqVDdH/qywE8EYBT8AXCdgYbYP27T/9iEsAoXtYFxkQBhJl+LBswylc7APa1IfanQzu8l1XXDuAJABzge/x0UL4Akh34WvXLMbx6Pqd+ATCd7cyZMwHCLNKTA7C1KXsHXVYH9fra7fpyAG8ZwJgDsPEy4Cb4CsBAF6DS6UrqF9MDb0VG+XIs5zFpIwdgtwPvdid2SC+3fh3AWwQwwASUmAdi+P7ZI/+/dgNAtbZf636G/Rf1y7rAHKfOB7BjANs03B94uZ1Udezh7tWhA3hLAAaQ2GKtbVaARA1bQHKsPB9kgrAQtupXnZJzsAMrfgt5zBsxsHWeh7vXqb1Ol1OnDuCRAYxKBa744+bgaNdsYNH1J4zfrwCM8mVnNbQUTFHXpTRwV3NviOV0TIfoftSVA3hEAOOhQEeKlalVpyy6g3pVhwPA3/vjj4Zd8O1Sv5wLXIFsDvLWxKG0PNyPTu71PN96dgCPBGCZHGSXJYzhGEwPhwNpdBJU7PMXP3wZfC2A7SQNDehxrgbi4jQ0wBdAb9LyTjnfTul1sz914wAeGMDBL/f46eClkJudJgVsTQ8AFLezkvqV3y8dFPhiUyYODcbxWw7AfT0h8Mq46qqrRt2vvvrq9RPAHODDNbfsU+aZsqN+rrjiiiO76oz/2U+fPr32sBk7v7VlN2Q+atPUcUOmvUlcDuCBAHzJ1nsirMvA2rwxfK0CRpWGNRqs6eHYqdWrXzhfBLDczqhw7MpMXwa6agCo7jgdKeB4oE/nlEIa69mzZ1cXL14cbb///vtDGgcHByvtwIS0S3kb6z/SBVg8adTuwG6s/MTxkj+VE6Hq5/HHH1/ZXXVG+Z47d2518803HznPxhF/5noogz51wDm15dYn/rg8+E5eyXNtupRFKp4pfnMADwBgwMjAGLZewJtaIF1gBIiX1OiltX6pdGt6SClgBt5wSxNsUdnYe/FuiL0niFsKm7QE4PjYmsYGCOnU29refPPN1csvvxzSFBRQbjV5HeoY0v3Od77TdMmUEWU1VB5y8ahMnnrqqdVrr73WlEd78I9//OMVZf3666+HeCjzF198cfXMM8+siBt4A22lp7DmRsOxNRvpc2zuWmt/p30A3paNshsi7do8lo5zAG8AYECI8uVVQYBXS0SmAGxNA5wjswGDbg/dfvNa+aYAHNzOrFo+OBHge9kA3uGU5ByAaagtqmPbAI47EVAgz9vqLJQNkGndAPbYeSR+4LjtDVgDaiANmLsgTD5rNuIcosyIo/VmxDVs+8aeg7ADuBHAwdRw/HRQo9hk7cI4NQBmIoaULHGx9oOgq9B6P/CZyhM4MTMwtRhFG3s2oKT5Lwdg+4qjXIOwv08NYHVklBkdTWVg8zjkZ9J4+umnlWxTOGb+iLtVlTdlvvLgZ599thOa5LVmGwLAferr0Ucf7byGIdtUV1wO4EoAB/AeOxUUmcAr31yBl1DqlxCzg0wPmCeIQ/ClYgCmtfumAMxav6hkjudcIAp82eNFdogP27JMDwp1vB3062oY/D8XANOh9diIva8m732OoUMDhj4b6pTz+6RbOoc45wBfykQ3wq781pTfpgCmHWD/btlQ8WPUUak8uv5zABcAHKB7kkGPE8FDAVOAdQmrBTD2Wmy1VsGl7L4xgC83PVx5ZOlKu3IaFY1JBFUs8CoUgGNgdzWOOQGYjobdcKzHRzo0ds++2xhmCGAxhdkhVwZzAjBl03KzxJTCObYPdrX/bfzvAI4ALJUKIPnMrLR4WrDAqzBWwNiD2TEHoFoBo61Mvj994dbVfzz+K2EXeBUCedK055EfwRWg/te3fu/IABzxc/yfPPyRLIBbZ8PNDcCC8BgdiThfeOGFHHuqfufmMGQHJ0+AYy7bXABMuWAOadmwX9OebT+cw2cH8OEkBh7zLXRZh1dAJLR2WYFXoQUw4OU1Qfj0opzjDqlBtxx8lZa118r0YBVtPAAnAPNGZR2nUAqY/6wJpKsBzhHAdLoxHiWHgB32Y+LpKtea/4lnTuqXcp8DgLnJ3XXXXS3sDR4eQ9VLTd21HLN3AEbVrk0Lh4uZA0sUaQzdHICtGULwBbzYfYPN9uBEEnQH11y7+o333rhWvkDYpiH4orpj9UvcFqjxAByVzjkxgAVfwl0BML2PwZShFA2dmvg23WSnbumAuWMBxlxsvyqXOQCYcsEUVbu99dZbszQ9qN53CsCCq0KAF3bU7SFssZvi9iXgSokqjIGo71YBpwDMgugo3pzCRPne+dM3HIFvCsBMRSaedQVFq5wBYWCaGlCbGsA09pa9thOljlPHUjltEg4JO9zmhnBxIk+tm3x5UeLagSY7j+zsX//614OpBXMLx2vnqQLgs3MjYcfGyg7wMIVMDWDKhPy2bExE4Qa7SfsY89ydATDgA65ADpX5ife8c/WHv/DuANq/vvd9q5d+5+dX//7YXev9u5+/+zIYpoAoVWoBLNMDpgaB1yrWuMJy8I3TQ4FjBtH5fNabLqxrGQC2EzDWxydswNtSwFJ/dJKWHWDxqN2iatQBh/A8wETUB3bKQxzWQEr1lQsBOLO1ajfKrqXMNzk2l2f9XluWrV4QPO0A05aNmwz5Ud7mGO4EgIHff/72HUf2Vz91bmV3C18+A+AYwlK7cWjhy6w0wMvMNJRqCbxUuMwOSlNKW6HSwh0NUKuRcEPBNIJpQ65sUr+5dR3ISzwIZwE85iCcAKz814bARkAAXi2bwFObVuo40u7r+5vKaytYUnlCsTFoVLthE52LyhsDwLpJtgxIDtE2UnUz9G87A+B/e+j2FfsrD9+x3rsALAgqFAxtaOGL/y9TjlGmOVODrSANuAn+MfBJV2mhBDGdcD4hah77svUlFoCD/TfyrOA88iVPCWsvFoTHdEPrC+Aj5dVj4GnTR0yAATSH3FCvm/grtwx+DmmKsXXR9/MYACZOTCQtGy6Fc7kplcrSAXzoCmZhKCgSYpMNapfXvx9OPS4VqP7j2D/4uZ85YvLIATheYJ1zNb1ZEzpkggCmdjad0iMEwLi+Cb4CtgCcshvb8+PPLSAYAsDhGhrtfNgE6aBx3mu+93Hmr4FAzYyxUv6mKPdSflr+GxrAxNc6QLpp+bdc76bH7hSArfrlc04Bl8wPmAJi6NaoXVsRgPAvf/3sGr45BQzgUzDFtIH6FXytCaK0qhnpMuMuB2Dr2mbzm/s8BQj0uFkDOo7ZRAHSuWt9f3n8pWPXbJs+/k5R7rk20Pr7kACmLVAWLdsQJqDWa97k+N0A8MkzwfxgTRAWvnwWBAllciAUcHH9knkB04HMAS2FyznYfBn0U/pKN6V+GTC0dl/SAva4ssmtLTZBoHABbSpfgNsCWMpXIfmjUafOTf02FQhawEjn5JG/j+cB6dTaFVHaLVNfeQTua4aYqtxTbaD1tyEBTFytrnjUUd9yb73WIY7fKwDjCYEyxTsCTwkKEJgBwb7QVSUAXoBqlbfgSxgDGNinlDUQZZDPKmDMD+yAtGTH5VwG6KSABV6FKc8J5T8VTgWCVi+APt4QrQNd2JoBAuq2ZttkBH6qck+1gdbfhgIw8bQuhcpgKue15nnK43cDwMdPZxUw0MWmut4NbPuo3Liyguo9tPfiiVED4HgNB8VJXHg+4GkRA1hQBbI6Pg4BrI4jFHgJU1OX4/Pj71OCgI5UOzjWxw5M/LXqSiYFzsGPtmbTOXGZ1nyfstxr8lc6hjKq2UqmAm7ADEq3bEONQZSubYz/9gPAxr1ryEJE9dJQ/uG+24ILnIVvyuzBzYD0c+AHrizAA3xTAC6ZH2RayAE457pWKo8pQUBHrnW6b+18rXZmDerwaNuyYA/uYX0eh1vKvQSyUt2O9d8QACYO6rRlox/2MUONVQ618e4UgAGgdtlgscfGdtbawskdF1TvyTPByyF2f1O6MYDDJIuCF0WsfmMAo2JLg2iYM/ByyAE4tXZE7vr0ewsIWiGoNHIhadeuhdCqNungtXEDAWtj5txaM0QfZU55cO0tI//kSTfgXHlu63fyUrPlbhyc3+qX3ccEta3y6EpnNwB8zbVrBTo2gFG9t/zk9au//7WzIc0aADPtOTdwpgpC/QJpZtmlFHCXgsXEUloHIrV2hNLOhVMCGOXYsugKHTd3HfHvHFtr3ohvLJxb6w3R10MD+/QDDzxQw7FwjOzN5G2InfT7Ap30azaVa5zfloFO0sGMRBxxHS/l++IBjHLk8UP21xjADLoNoYABL2n9xbl3r/71tz643rsAzLoTwLHUIFCv+Bpj+xWArQsaqrbLhxfAl3yAS4N3ubxNCWAAAIRrNzphDTSIs6WTx+qq9XxA2johoI8NFNgDI3YmLfTZUezMRkR9C4xcb659pH7nvJqNpwjKFrWrne+1N0alQdku0fSgstsJAON9kANwUJ/XXNvUiFQ4hICX/Y/ef9Pqnz/6vrDXABh7L4NtNfBH/WrN4RjAeD90qd+Qzw4PiNzAn73W+POUAL50TXWdmc5Y634EIGp9f4k3BXZ+qwVFXzMEadS6yAlGQ4fAHJs3eYnbRu47x25r040il5cl/L4TAGbhHQtga4fF5awGgraygo03AV4AbOHLZ+t7jPombZkciMfGm/pM3vBBjldYkwIGwKnJGnFcQDxn/8V+3OqCRvxzAHAthGoHvAAEarFmy8GTOGrXrSD/LQBTvXIOAJzDJhOH8lYKyfc2NhR+n3It5X2K/xYPYNQpU35zAAbONSCk8AVebLwo3n86f2vYpXxjAMv8QKj0U5MrShVLmsyIywGYiRXAtRQHJgzc13IA7jMAR3pzAHAtLGvWhOBxtWWRm5z5gEfeltXKat4mHNcvZd8yEDc29LgZ1ABvGwDWTa3G5BSX69y+7wSAsfPK9isVKhUMEGsADMjPv+Ntq+c+ePMavLUABr5BaRe8HFIVj92WgTct+KNlLq0HRI3pABtzaRW0PgNw5HcOAK71OMjB0pY7cKhd1AXwl4DDf7VmiL5qjTRqr39sABN/jRImz2Nv1DVt09btUj8vH8AnzwQ/XGsKEHwJa+GbAm8XgAEvPsAoZgDe0ghQrQzQaeEfKWB5QDAVmTccd3lPkCYK2U5BRgnbSRhdA3i5fO8SgFFLLXCQ72+ubIir1gwBkDg+F1fudxR76xq4Y8Ova4p1Sxn3zWs8MJorvyX8vnwAH86CyyngLg8EKglIP3TT2y9TviUA44aGeQMbbg3kbWPgeHYL3xjA2IBr4Eu8ABhbcc4E0WXCsHmzn+cA4KFMEICBjlu7Wd9fWyb63GqGYFptH9VGvlun5NZeY5/jutT8NgBMvqmfVg8N1d2cwsUDGJDJ/ioISwFrGnJNgaNgBdw4jG3An7vlxqalKeP0ZXrIARj1y4I8qOT43Pg7x8QTMKwCrvGgiOPU9zkAuBYSNcqsdnaVfFRVDrkQ2NTGWWtDTaVFOjx2196Masus73HkJ2d/3RaAMf+QVqq8lvTbogEMfFlUJwfg4ANcaRoAwPj4luAbwMtaEpVxphpCmDBx4WxYkS0H4JqBN8UNzGP/XwvgvvZf4l8SgEuKFaXUMmiG4qRz1+wtqpr4cuBSfeZCzBGcT3q10O8L2K7zSmqePG5rY10O0suV2RJ+XzSAAWGXB0SNiqSigDkDdjkAY3KoMWeUKl1paDlMC2A7AKf8lOLSf7iXlcwPpenLiiMXTgng1kf8Etz4r3YRHeCB+1ntXjuoR7zAs48ZwtYP16Id1ztgiC0am3WfnYG1Vne30mAceavZGFxkAgZ51873WrdDpVHj/WLLb26flw3gk2dWz//SLUd8cWV+IIzfNNFV+AD9pTtvOQJha35gxh0Q7Yon9T/nsdtlKS2AsQEzEy681r5j5pzi7zI/fO8bn+nl/6v4pwRwq8sYHV/5jkP+m8Pj+yZmiPia+I6yp44E5E3ClieE0nWQh5pNJoQ4z5haWjZ5q/R9skiV6zZ/WzaAC8tQAuDaQSwVOAC2/r9Sw4IwJoi+AEY9Y5MWgAVfQg3AsRJay4CZzA85BVxaPU3XXAqnBDAds9bLADXF8alrAeStnboFAK3Hks85woJ81T4llGzkxFOzCcBxnXE+Crtl6xoYjNOY0/fFAhgQlqYg91kFjTjxBRZ4CQVfQtR2H/svoCQ/9m0cMYBZhL35hnHsVDA/5ABcM4Ou1BinBnDto3EXEFrMBC0dv8+xmCGATKncp/ivpa5z8CTfmwJYcZBGy7bUacmLBbDUqvX/tZMw+kxBDpUfeUNYADP1uHVaM8qXwUC9HSNl/0UBN5tLjp9e8Zp54JsC8KbmB8qipVOWIEhcrXttR6aT5qYMozRb4mnp8H2PHbqcWss1dzx1XTvzbmwAY1rB3Ne6cW1LW5hnuQBOTMCwAG4FmhomYLeTMmIA186sIz7gy40Ac0j8WiIpYOAb3rrcuGg8g2+4mOUAHLwfKm3JuvY4nArALenSSXOqEvjy39y2OS4ePicAh77T423Ic725xf3Kfl8kgDEVaAnKnALu67EAgLH1xuYHQIwCxuuixgxB+vLQEIBTJogvf+yOZtODBt8EX6uANQOOSt7U1tgCwiEbP+BsmYKbWwmNeMjX3DZG+8mb7YhTf54bgCkPyoinm5aNcYO5lW2pbhcJYAAouFkAAzr2Fv/fuHCAO1OLcwDGV7gLwBa+UuUpBRzektGx0E6cv0sN89LUYwE4ngHXd/GdOK0pAEznAVC1mxZmifPe6sZWm94Qx5Ue4ePr2Nb3OQKYa6c9tNyMqR8m5TD4uq2y2ySdZQL4+OnL1n8Q6AAwkzMAad+CAbAaiJMJQstQBn/ggrkA+OJJockhuilYAGN+CPDNvF6+lG+ui8E1wdeqX03A6Lv2Q5zutgFMZ2tZrYzOlhsBJ67aN1cQDyDfdG+Bc2niSFwP2/jeUtelGwjlXrOV4rDXC0gBastW8oqxcc/h8+IALIUK4Kz6tQDua35QhQBg2YFjAJcG4kgX9W3zlgLwq18433sqsxbeyQH40tTjK3vffFQGhC2dUiYIzB61OyqVDkanZa91O7OdMbfUI/G1+P4qD5uELfbmrsV+bD1s4zPXXet1UoIn8dRspTji6yXOlqci0s8NzMZxT/19cQAGjlKYKQAH80NBodYUuNKwZggpYOAaD8RxU+AcvR2ZfClvFsB4QABfjmevyYs9hnOkfn/43CfXKlgmCOy/eEbUzv6zcac+9wEwI9g1O/DFdgtA8T9tgaU6eM78ANRbVhEborO2elzUqjTKkuvRbsuWMmTXDS9VhzW/AbiWJw/dbFNxjwFg0iHe2huE2gfXRBtO5XMuvy0PwCfPBLhZyKF+2YHjpuYHKgbQsTqa1K8G4IAwadhF3jkWm3EMXwtgzA8yQXA8e58GIPVr4WtNEG9885GNZr7FeWoBsBr9NsPcgAudtWXwhim9AC6+/tbvrZDIDR4qXWBL3nAPAyZMKOE7O4/lnM9gNMdTV6TfZ29R7tRv6YZF+jVbiwLm+nSD46Zbu+kGzbkq07mFiwIw4NLrh2IAS2luan6ggkgHlZsDMAocxcuxKeVr80a+gC+z4DaBrzwfgG8KwEOrX65t7gCms6c6Vy0E6Mi1SrSm41Jetb60pF1aUyG0rUY7dg2YYjt3zTnxMbkbn/IcH5/63gpgtceWJxvSLU2brqnTsY9ZFICBK0pTgJPKlP03TL7YYKUyW9iANQVgVLCdEcdxmD1snvRZ+QrLYvZYN/hIfg5OrPSeuBSAh1a/pD1nAGvFMltGffJcgkkcd833IeFPXH3s4ingDfkbChx1niqP2uvvA2DSI/6WwVWuO+cnnsr/tn9bDIClSjXAZSEn0FF4HDdEIZYAzE1AM+I4zrrEKV+E5CtMiW58VVGcf6Yo88oh1gmOFbA8H/q8dj5OJ/4+VwCXVA0dtMX3l2tMqei4LGq/k36r+aMEs7kBWIvf5MpjbACTLmkA8JYNr5NcOeeuZRu/LwbAJfW7Bt2Gg2+2wAErLmfW/quBOAtggB+bRYAvN4owILjhbDTi571wFr5WAQPgIT0fbBnMEcAyGaSgyYBUyxTWEshtObR8bh0ALJkhAM3cANz1xLANALfWM6Duq7hb6r7PsYsAcJf6BXaxZ0KfwrDnAGBMDSkAA1jZmlMAJj9hxtyG8CU/duBN6jcG8CZr/tprjj/PDcDqRHTAOK+Xyqrt8ZSBrSEG3+K8ACFuFDVbSVHODcClvKoMtgFg1XWLvZ26mOMC7osAcEr96hEf9RsU6QCwUyMKFXz4howYwFK3AJrjUjcHbgb638bZ+hnTwxfv/UBQv9b+KwCjfsOSkz1m09XkZU4AllLMwZfrGQp8NWVTOoZ81C7tCBhyi4rPDcBdr31SHdTceHQzLZVj13+UT4u5p1TWXWmN9f/sAQzg9Noh4Gd34DuG+qWwASjeDjkA86hL3thxQyMf3Aj026YVRjwsKITpIQVfTcRoXcKyJV9zADBmAjo+na2U99ZH/zEnQmBrxEWsdsu5dc0FwHhOMPDWVQeh34zkhpare/JU+7RBfUjFp0xYuTTG/H32AM6pXyngodQvwLMTGLoAbFdbw0YbXNM2HGyzFY3pgTWCSwAeY+DN5mHbAKajo4wAEiPXQJUOVjN4wnEt6/6OPRWY/HAttRvH27LXZ35HIWN7pVwYYAQi29p0c6g11ZDfmk22fF1n37D1ZkfexrD9983/rAEMBHMeBjIFbGr7BbxAnvCJC2fXECZtVkXLKeA4XY7vWwnxecCXt2MIvikF3PLizjj+lu8oBXYe/cfY6UCAno4b760qhbjiOHLfa6DeUk6pY7kuwNW1l8wqxKsbUXwtlA/KFFs2bnlM1wXUmGsAJzckYAO02bkhAD52IM5u/YL5znGcKzc/0mypB8pVE0V4cintXdedKtPUb5QPTxyltOx/5G+otFP5aflttgAGiFyINTnEn+Xi1XLBHBugy9uNj50KU3ufv/jhFQvkvPLkPeulIVsB3JqH3PHW7vvGVz8ezA8pAKPWWzpGLr2a30ln7L0mHzXH1OazJq4hjhkzP8QN8ABQ7iYWQ7v2O/ERf58yIE9d+9AAJL6uNO3/fa5rjHNmC2BUqV64GYNX6hfbq0DdVTgWuhyL2mVdBi2MPgcAc80XbrspvJwT+ArAsffDUKuddZWZ/9/+Jg8vMy+zljYwSwCjPlO+tQIxA141i+5Y6DJoBnRZBtJC137uUsA2/dgE0VLoqWNRtNiVBV6Fsfp99rG7m17cmUrLf3NIeBuYRxuYHYCBJrtglwtjm2uA7TXXhhlqKElmquE98fSFW8M6DHoXG6GFrv08FYDJOx3iB195MOyCb6yAL024ONHrsdA73Dw6nNeD14NtA7MDMPDEsyEHXtQv6liQJQRgKFLWgkAZs8JTVmkAAAXKSURBVPaCXYHMwjcHYN7NFgOYAUA7CKc8Den6Rt5Rv3g8lACMEuZYwdpWon/2Tu1tYJltYFYARtWWvB4EQI4BtIBaq6ApFHgVfvfzd69qAIwStm+pIC+4lmn6MaHSB8At9udc5wCmKPUSfGWC2PQV87k8+O/L7Lheb7tRb7MBMDBCxQI3ga4UMgmDXeBVKPASAt9aAKOA8YZAUdO4ATDvf8sBeFMlyvlB+T55z1r5ooBj8wOuaPj7Kl/e8Xaj43k9ej0GzsyhIQAj9hJw4/9qARyr39gEgScEbyZmAMzOKgPAeGHEANbsO62G1qf8BG9MHoBfpocYwMCXqcg2X33S83O8s3sbmGcbmIUCRt2V7L45+NYo4BjAmBqALmqXQTrgBkwFRTVUAMxqaDGAyQvp9gUw5wF7wKs9BWDg6x4P8+w0aiMeev1s2gYmBzDwxZ5ba3oQAFsVMANzeESsodsxbTgGsG4CpBumP/dY+pJrxc9X4FUYAxj4Pv9p5t67x8OmDdzPd0jOuQ1MCmAgV/L3FfTiUPAllO2X0Np/AS5gf+j2my/ZdDNKN1c5KNWc+g0z8BoBjNLGDxnoosIFX0ILYODLNGSHr4Mj1zb9991pG5MBmEd+DbqhfvvsABjQAkTULd4RxInSDPuhaSE2L3Q1YI5n4kYKwNwMADs3j654+J+4gK+mO8fwtQD+0dcuuvK9bnc6V0378GP2u74nBTDqF2jW7BzLDmBxAUOhWtACRLwKWmGb6gDEcf4db0sCmBtF7bvnZO+Npzxb9SsAO3z3uyOm2qH/tvttYjIA07gAXVCIzGDTLDZ9jkIdq3DMxkleuCmkFDAA5kZAPnJ5CNd0aHJA8cZ7DGDgG7wd3OabLdNcWfvvuw+pXa7jSQE814IFwCUf4NIkDFQv6wNjGold3gRiC2Dsv5/40Lvc1cxND37z2cM24ABOVDoALrqgJV5/JNWLecK6vgm6NrQAZoYbppS53ow8X64wvQ2M1wYcwBkAp9aAYAAudkETeHFvk+qtATCTMGjY2K29gY/XwL1svWzn3AYcwBGAASoDfSn7r/WACOA9fmkJSbwwNPXZwjdlgkD9MvMONzPimHPj8Lw5vLwNjNsGHMARgDE/lBbhYQAOtzLswAKv4Fuz7oTbe8dt0A4ML98ltQEHcALAOfsvHhCoY8CrCSCa/NG18A8rraF43eTggFgSIDyv47ZXB3ACwNb+iylCM/E0A0/wjWffxQpYA2/MxkM1u8lh3MbssPDyXVobcAAbAANITAx9ARzbf5n9huLFNW1pDcPz6zDzNjB+G3AAGwDL/zcFYKlfu/6EzA+yAQvAeEOEKdHHTjl4Tfl6hx6/Q3sZL6uMHcAGECjVGL4yQXQBGPhignBzw7I6gAPL62vKNuAAPgQw5gfWf4gBLPsvoVW/sf03rA9x7JQPspkb2pQN29N2sC6hDTiAD4FRMj8IwjGAMTXILc3tvN7hl9DhPY/zaqcOYANg1G9OAcsEgSsabmha2N3dyubVoB0wXh9LagMO4MNV2R666e0BvikAA10AjJmBhXZYu8FdyryjL6mje17n2V4dwIdvQOYFnDF8GYBjVbTg0cAi79dc6+B1G697tngbGKwNOIAPFbA1PwBdBuTCgu8O3cEam6uweaowr5fp6mXvAYwpAfPDWunypg2HrkPXVZ63gS20gb0HMHd/IOzQnU4FuALzst/XNuAA3sJdbl8bl1+3g9XbQLkNOIAdwP6o6W3A28BEbcABPFHBuzIoKwMvHy+ffWgDDmAHsKsfbwPeBiZqAw7giQp+H+7ufo2uYr0NlNuAA9gB7OrH24C3gYnagAN4ooJ3ZVBWBl4+Xj770AYcwA5gVz/eBrwNTNQGHMATFfw+3N39Gl3FehsotwEHsAPY1Y+3AW8DE7UBB/BEBe/KoKwMvHy8fPahDTiAHcCufrwNeBuYqA04gCcq+H24u/s1uor1NlBuA/8HnH+QJpRUfv8AAAAASUVORK5CYII="}}},{"cell_type":"markdown","source":"### Why do I choose to use Dask?\n\nDask enables natural scaling of Pandas, Scikit-Learn, and Numpy workflows with little rewriting when dealing with large datasets with a single computer. It works effectively with these tools, copying the majority of their API and internal data structures. Furthermore, Dask collaborates with these libraries to guarantee that they progress in a consistent manner, reducing friction when moving from a local laptop to a multi-core workstation, and finally to a distributed cluster. ","metadata":{}},{"cell_type":"markdown","source":"### Learning Outcome\n\n* Analysing images on a single desktop using dask\n* Preprocessing of image dataset","metadata":{}},{"cell_type":"markdown","source":"### Importing the libraries","metadata":{}},{"cell_type":"code","source":"!pip install \"dask[complete]\"","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:17:31.507907Z","iopub.execute_input":"2021-05-31T00:17:31.508422Z","iopub.status.idle":"2021-05-31T00:17:38.259972Z","shell.execute_reply.started":"2021-05-31T00:17:31.508382Z","shell.execute_reply":"2021-05-31T00:17:38.258425Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"!pip install dask_image\nimport dask_image.imread\nimport dask_image.ndfilters\nimport dask_image.ndmeasure\nimport dask.array as da","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:17:38.265169Z","iopub.execute_input":"2021-05-31T00:17:38.265592Z","iopub.status.idle":"2021-05-31T00:17:44.862997Z","shell.execute_reply.started":"2021-05-31T00:17:38.265556Z","shell.execute_reply":"2021-05-31T00:17:44.862038Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"!pip install scikit-image --upgrade-strategy only-if-needed","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:17:44.864654Z","iopub.execute_input":"2021-05-31T00:17:44.864996Z","iopub.status.idle":"2021-05-31T00:17:51.383252Z","shell.execute_reply.started":"2021-05-31T00:17:44.864964Z","shell.execute_reply":"2021-05-31T00:17:51.382082Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from keras.models import Model, Sequential\nfrom keras.layers import Activation, Dense, BatchNormalization, Dropout, Conv2D, Conv2DTranspose, MaxPooling2D, UpSampling2D, Input, Reshape\nfrom keras import backend as K\nfrom keras.optimizers import Adam, SGD\nimport tensorflow as tf\nimport numpy as np\nimport pandas as pd\nimport glob\nimport gc\nimport cv2\nimport PIL\nfrom PIL import Image\nimport matplotlib.pyplot as plt\nimport cv2\n%matplotlib inline\n\nimport os\nfrom skimage import data, io\n\nfrom sklearn.model_selection import train_test_split\nfrom warnings import filterwarnings\n\nfilterwarnings('ignore')\nplt.rcParams[\"axes.grid\"] = False\nnp.random.seed(101)","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:17:51.385356Z","iopub.execute_input":"2021-05-31T00:17:51.385678Z","iopub.status.idle":"2021-05-31T00:17:51.399174Z","shell.execute_reply.started":"2021-05-31T00:17:51.385643Z","shell.execute_reply":"2021-05-31T00:17:51.397999Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"!mkdir temp\n!mkdir temp-test\n!mkdir temp-train","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:17:51.40062Z","iopub.execute_input":"2021-05-31T00:17:51.401068Z","iopub.status.idle":"2021-05-31T00:17:53.588675Z","shell.execute_reply.started":"2021-05-31T00:17:51.401032Z","shell.execute_reply":"2021-05-31T00:17:53.587215Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"### About the dataset\n\nThe datasets was publicly made available by Kaggle but provided by the International Skin Imaging Collaboration (ISIC), funded by the International Society for Digital Imaging of the Skin, is an international initiative to improve melanoma diagnosis (ISDIS). The ISIC Archive houses the world's largest collection of high-resolution dermoscopic photographs of skin lesions. \nThe datasets can be downloaded from this link\n[https://www.kaggle.com/c/siim-isic-melanoma-classification](http://)\n","metadata":{}},{"cell_type":"code","source":"filename_pattern = os.path.join('temp', '/kaggle/input/siic-melanoma-384x384-test/train/ISIC_*.jpg')\nisic_images = dask_image.imread.imread(filename_pattern)\nprint(isic_images)\n\nfilename_pattern1 = os.path.join('temp-test', '/kaggle/input/siic-melanoma-384x384-test/test/ISIC_*.jpg')\nisic_test = dask_image.imread.imread(filename_pattern1)\nprint(isic_test)\n\nfilename_pattern2 = os.path.join('temp-train', '/kaggle/input/siic-melanoma-384x384-test/train/ISIC_*.jpg')\nisic_train = dask_image.imread.imread(filename_pattern2)\nprint(isic_train)","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:17:53.590603Z","iopub.execute_input":"2021-05-31T00:17:53.591003Z","iopub.status.idle":"2021-05-31T00:17:58.091972Z","shell.execute_reply.started":"2021-05-31T00:17:53.590956Z","shell.execute_reply":"2021-05-31T00:17:58.090867Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def grayscale(rgb):\n    result = ((rgb[..., 0] * 0.2125) +\n              (rgb[..., 1] * 0.7154) +\n              (rgb[..., 2] * 0.0721))\n    return result","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:17:58.093275Z","iopub.execute_input":"2021-05-31T00:17:58.093579Z","iopub.status.idle":"2021-05-31T00:17:58.098953Z","shell.execute_reply.started":"2021-05-31T00:17:58.093549Z","shell.execute_reply":"2021-05-31T00:17:58.097775Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"**Taking a bitchunk of the images and analysing it**","metadata":{}},{"cell_type":"markdown","source":"Splitting one of the isic image into 4 non-overlapping tiles for segmentation purpose\n","metadata":{}},{"cell_type":"code","source":"from PIL import Image\npixels = []\ni = Image.open(\"/kaggle/input/siic-melanoma-384x384-test/test/ISIC_0546632.jpg\").convert(\"LA\") \np = i.load()\nfor b in range(4):\n    block = []\n    for x in range(32):\n        for y in range(32):\n            block.append(p[x,y][0])\n            \n            \n   ","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:17:58.101781Z","iopub.execute_input":"2021-05-31T00:17:58.102123Z","iopub.status.idle":"2021-05-31T00:17:58.135161Z","shell.execute_reply.started":"2021-05-31T00:17:58.102092Z","shell.execute_reply":"2021-05-31T00:17:58.134147Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from PIL import Image\n  \ndef main():\n    try:\n        #Relative Path\n        img = Image.open(\"/kaggle/input/siic-melanoma-384x384-test/test/ISIC_0546632.jpg\") \n          \n        #splitting the image\n        print (img.split)\n    except IOError:\n        pass\n  \nif __name__ == \"__main__\":\n    main()","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:17:58.137387Z","iopub.execute_input":"2021-05-31T00:17:58.13783Z","iopub.status.idle":"2021-05-31T00:17:58.144429Z","shell.execute_reply.started":"2021-05-31T00:17:58.137786Z","shell.execute_reply":"2021-05-31T00:17:58.14369Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from PIL import Image\n  \n# opening a multiband image (RGB specifically)\nim = Image.open(r\"/kaggle/input/siic-melanoma-384x384-test/test/ISIC_0546632.jpg\")\n  \n# split() method\n# this will split the image in individual bands\n# and return a tuple\nim1 = Image.Image.split(im)\n  \n# showing each band\nim1[0].show()\nim1[1].show()\nim1[2].show()","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:17:58.145372Z","iopub.execute_input":"2021-05-31T00:17:58.145617Z","iopub.status.idle":"2021-05-31T00:17:58.312639Z","shell.execute_reply.started":"2021-05-31T00:17:58.145593Z","shell.execute_reply":"2021-05-31T00:17:58.311498Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"fig, ax = plt.subplots(nrows=2, ncols=2)\nax[0,0].imshow(im1[0])\nax[0,1].imshow(im1[1])\nax[1,0].imshow(im1[2])\nplt.show()\n","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:17:58.314616Z","iopub.execute_input":"2021-05-31T00:17:58.315386Z","iopub.status.idle":"2021-05-31T00:17:58.886992Z","shell.execute_reply.started":"2021-05-31T00:17:58.315332Z","shell.execute_reply":"2021-05-31T00:17:58.885915Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from skimage import data\nfrom skimage import color\nfrom skimage.util import view_as_blocks\n\n# get astronaut from skimage.data in grayscale\nl = color.rgb2gray(data.astronaut())\n\n# size of blocks\nblock_shape = (4, 4)\n\n# see astronaut as a matrix of blocks (of shape block_shape)\nview = view_as_blocks(l, block_shape)\n\n# collapse the last two dimensions in one\nflatten_view = view.reshape(view.shape[0], view.shape[1], -1)\n\nprint(flatten_view.shape)","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:17:58.888404Z","iopub.execute_input":"2021-05-31T00:17:58.888732Z","iopub.status.idle":"2021-05-31T00:17:58.914229Z","shell.execute_reply.started":"2021-05-31T00:17:58.888703Z","shell.execute_reply":"2021-05-31T00:17:58.913128Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"!mkdir temp-tiles1","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:17:58.915402Z","iopub.execute_input":"2021-05-31T00:17:58.91569Z","iopub.status.idle":"2021-05-31T00:17:59.652619Z","shell.execute_reply.started":"2021-05-31T00:17:58.915662Z","shell.execute_reply":"2021-05-31T00:17:59.651415Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"mkdir temp-tiles","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:17:59.654041Z","iopub.execute_input":"2021-05-31T00:17:59.654364Z","iopub.status.idle":"2021-05-31T00:18:00.391708Z","shell.execute_reply.started":"2021-05-31T00:17:59.654333Z","shell.execute_reply":"2021-05-31T00:18:00.390304Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"!ls temp-tiles","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:18:00.393466Z","iopub.execute_input":"2021-05-31T00:18:00.393795Z","iopub.status.idle":"2021-05-31T00:18:01.136559Z","shell.execute_reply.started":"2021-05-31T00:18:00.393751Z","shell.execute_reply":"2021-05-31T00:18:01.13517Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"fig, ax = plt.subplots(nrows=2, ncols=2)\nax[0,0].imshow(isic_images[0])\nax[0,1].imshow(isic_images[1])\nax[1,0].imshow(isic_images[2])\nax[1,1].imshow(isic_images[3])\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:18:01.138513Z","iopub.execute_input":"2021-05-31T00:18:01.138964Z","iopub.status.idle":"2021-05-31T00:18:07.790309Z","shell.execute_reply.started":"2021-05-31T00:18:01.138917Z","shell.execute_reply":"2021-05-31T00:18:07.789181Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def grayscale(rgb):\n    result = ((rgb[..., 0] * 0.2125) +\n              (rgb[..., 1] * 0.7154) +\n              (rgb[..., 2] * 0.0721))\n    return result","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:18:07.791878Z","iopub.execute_input":"2021-05-31T00:18:07.792178Z","iopub.status.idle":"2021-05-31T00:18:07.796932Z","shell.execute_reply.started":"2021-05-31T00:18:07.792148Z","shell.execute_reply":"2021-05-31T00:18:07.79583Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import os\nfilename = os.path.join('temp-tiles', '/kaggle/input/siic-melanoma-384x384-test/test/ISIC_0546632.jpg')\nprint(filename)","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:18:07.798049Z","iopub.execute_input":"2021-05-31T00:18:07.798357Z","iopub.status.idle":"2021-05-31T00:18:07.814144Z","shell.execute_reply.started":"2021-05-31T00:18:07.798331Z","shell.execute_reply":"2021-05-31T00:18:07.813372Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"isic_test = dask_image.imread.imread(filename)\nprint(isic_test)\nplt.imshow(isic_test[0, ...])","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:18:07.815334Z","iopub.execute_input":"2021-05-31T00:18:07.815801Z","iopub.status.idle":"2021-05-31T00:18:08.084032Z","shell.execute_reply.started":"2021-05-31T00:18:07.815769Z","shell.execute_reply":"2021-05-31T00:18:08.082941Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"single_image_result = grayscale(isic_test)\nprint(single_image_result)\nsingle_image_result.visualize()","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:18:08.085327Z","iopub.execute_input":"2021-05-31T00:18:08.085639Z","iopub.status.idle":"2021-05-31T00:18:08.492642Z","shell.execute_reply.started":"2021-05-31T00:18:08.085609Z","shell.execute_reply":"2021-05-31T00:18:08.491385Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"print(\"Original image dimensions: \", isic_test.shape)\nprint(\"Processed image dimensions:\", single_image_result.shape)\n\nfig, (ax0, ax1) = plt.subplots(nrows=1, ncols=2)\nax0.imshow(isic_test[0, ...])            # display the first (and only) frame of the image\nax1.imshow(single_image_result[0, ...], cmap='gray')  # display the first (and only) frame of the image\n\n# Subplot headings\nax0.set_title('Original image')\nax1.set_title('Processed image')\n\n# Don't display axes\nax0.axis('off')\nax1.axis('off')\n\n# Display images\nplt.show(fig)","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:18:08.494634Z","iopub.execute_input":"2021-05-31T00:18:08.495107Z","iopub.status.idle":"2021-05-31T00:18:08.816161Z","shell.execute_reply.started":"2021-05-31T00:18:08.495052Z","shell.execute_reply":"2021-05-31T00:18:08.815171Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"> We could notice the difference in the dimension by placing the images side by side ","metadata":{}},{"cell_type":"markdown","source":"Preprocessing of images is like data cleaning for tabular data.\n\n","metadata":{}},{"cell_type":"code","source":"import cv2\nimport numpy as np\nfrom matplotlib import pyplot as plt\n\nimg = cv2.imread('/kaggle/input/siic-melanoma-384x384-test/test/ISIC_0546632.jpg')\n\nblur = cv2.GaussianBlur(img,(5,5),0)\n\nplt.subplot(121),plt.imshow(img),plt.title('Original')\nplt.xticks([]), plt.yticks([])\nplt.subplot(122),plt.imshow(blur),plt.title('Blurred')\nplt.xticks([]), plt.yticks([])\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:18:08.817422Z","iopub.execute_input":"2021-05-31T00:18:08.817736Z","iopub.status.idle":"2021-05-31T00:18:09.100183Z","shell.execute_reply.started":"2021-05-31T00:18:08.817705Z","shell.execute_reply":"2021-05-31T00:18:09.098964Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"img = cv2.imread('/kaggle/input/siic-melanoma-384x384-test/test/ISIC_0546632.jpg')\n\nblur = cv2.GaussianBlur(img,(5,5),0)\n\nplt.subplot(121),plt.imshow(img),plt.title('Original')\nplt.xticks([]), plt.yticks([])\nplt.subplot(122),plt.imshow(blur),plt.title('Blurred')\nplt.xticks([]), plt.yticks([])\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:18:09.104582Z","iopub.execute_input":"2021-05-31T00:18:09.104927Z","iopub.status.idle":"2021-05-31T00:18:09.356315Z","shell.execute_reply.started":"2021-05-31T00:18:09.104896Z","shell.execute_reply":"2021-05-31T00:18:09.355222Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"plt.figure(figsize=(12,6))\nplt.subplot(1,2,1)\nplt.imshow(isic_test[0])\nplt.xlabel(\"Dimensions: \"+str(np.array(isic_test[0]).shape))\nplt.subplot(1,2,2)\nplt.imshow(isic_train[0], plt.cm.binary_r)\nplt.xlabel(\"Dimensions: \"+str(np.array(isic_train[0]).shape))\nplt.show()","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:18:09.358122Z","iopub.execute_input":"2021-05-31T00:18:09.358534Z","iopub.status.idle":"2021-05-31T00:18:13.078829Z","shell.execute_reply.started":"2021-05-31T00:18:09.358492Z","shell.execute_reply":"2021-05-31T00:18:13.077798Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"!pip install dask_ml","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:18:13.080141Z","iopub.execute_input":"2021-05-31T00:18:13.080458Z","iopub.status.idle":"2021-05-31T00:18:22.208538Z","shell.execute_reply.started":"2021-05-31T00:18:13.080429Z","shell.execute_reply":"2021-05-31T00:18:22.207223Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import dask_ml.datasets\nimport dask_ml.cluster\nimport matplotlib.pyplot as plt","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:18:22.210303Z","iopub.execute_input":"2021-05-31T00:18:22.210612Z","iopub.status.idle":"2021-05-31T00:18:24.184421Z","shell.execute_reply.started":"2021-05-31T00:18:22.210579Z","shell.execute_reply":"2021-05-31T00:18:24.183038Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from dask.distributed import Client\nclient = Client(n_workers=4, threads_per_worker=1)\nclient","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:18:24.186037Z","iopub.execute_input":"2021-05-31T00:18:24.186372Z","iopub.status.idle":"2021-05-31T00:18:25.74983Z","shell.execute_reply.started":"2021-05-31T00:18:24.186342Z","shell.execute_reply":"2021-05-31T00:18:25.748618Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import dask.array as da\nfrom dask_ml.datasets import make_classification\n\n\nn, d = 100000, 100\n\nX, y = make_classification(n_samples=n, n_features=d,\n                           chunks=n // 10, flip_y=0.2)\nX","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:18:25.752228Z","iopub.execute_input":"2021-05-31T00:18:25.752544Z","iopub.status.idle":"2021-05-31T00:18:25.802255Z","shell.execute_reply.started":"2021-05-31T00:18:25.752511Z","shell.execute_reply":"2021-05-31T00:18:25.800824Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from dask_ml.model_selection import train_test_split\nX_train, X_test, y_train, y_test = train_test_split(X, y)\nX_train","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:18:25.804084Z","iopub.execute_input":"2021-05-31T00:18:25.804556Z","iopub.status.idle":"2021-05-31T00:18:25.86184Z","shell.execute_reply.started":"2021-05-31T00:18:25.80451Z","shell.execute_reply":"2021-05-31T00:18:25.860861Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"threshold_value = 0.75 * da.max(isic_test).compute()\nprint(threshold_value)","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:18:25.863156Z","iopub.execute_input":"2021-05-31T00:18:25.863775Z","iopub.status.idle":"2021-05-31T00:18:26.940845Z","shell.execute_reply.started":"2021-05-31T00:18:25.863718Z","shell.execute_reply":"2021-05-31T00:18:26.937592Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"label_image, num_labels = dask_image.ndmeasure.label(threshold_image)","metadata":{"execution":{"iopub.status.busy":"2021-05-31T00:23:14.387247Z","iopub.execute_input":"2021-05-31T00:23:14.387603Z","iopub.status.idle":"2021-05-31T00:23:14.413978Z","shell.execute_reply.started":"2021-05-31T00:23:14.387574Z","shell.execute_reply":"2021-05-31T00:23:14.412612Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"### **References**\n\n* https://docs.dask.org/en/latest/why.html\n* https://examples.dask.org/applications/image-processing.html#Setting-up-your-environment\n* https://www.kaggle.com/c/siim-isic-melanoma-classification","metadata":{}},{"cell_type":"markdown","source":"","metadata":{}}]}