Benchmark - Max.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Benchmark of methods for brain tumor segmentation "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import warnings\n",
    "warnings.filterwarnings('ignore')\n",
    "\n",
    "import nibabel as nib\n",
    "from math import sqrt\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import scipy.stats\n",
    "import numpy as np\n",
    "import cv2 as cv\n",
    "from sklearn.cluster import KMeans\n",
    "from sklearn.mixture import GaussianMixture\n",
    "from skfuzzy.cluster import cmeans\n",
    "from dipy.segment.mask import median_otsu\n",
    "from time import time\n",
    "import glob\n",
    "from tqdm import tqdm_notebook as tqdm\n",
    "%matplotlib inline\n",
    "\n",
    "N_ITERS = 10"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Extracting and preprocessing "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "44\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e12265a5bbde4a97996b226a8404d2be",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Loading images', max=44), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "32c54457810d42a4bfe66d475b589710",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Preprocessing', max=44), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "NUMPASS = 5\n",
    "MEDIAN_RADIUS = 11\n",
    "HIT = 0.25\n",
    "BLUR_RADIUS = 3\n",
    "\n",
    "\n",
    "N_CLUSTERS = 5\n",
    "N_COMPONENTS = 2\n",
    "\n",
    "M = 2.0\n",
    "EPS = 0.01\n",
    "MAX_IT = 100\n",
    "    \n",
    "K1 = np.ones((3, 3), np.uint16)\n",
    "\n",
    "IMG_ROOT = '/Users/macbook/Documents/Education/4_Year/DiplomaWork/Images/brain/TCIA Low grade glioma/nifti/'\n",
    "\n",
    "def get_data(root, patients):\n",
    "    for patient in tqdm(patients, desc='Loading images'):\n",
    "        img_path = root + 'LGG-%s/LGG-%s_T2.nii.gz' % (patient[0], patient[0])\n",
    "        seg_path = root + 'LGG-%s/LGG-%s-Segmentation.nii.gz' % (patient[0], patient[0])\n",
    "        img = nib.load(img_path).get_data()[:, :, patient[1]]\n",
    "        seg = nib.load(seg_path).get_data()[:, :, patient[1]]\n",
    "        yield (img, seg)\n",
    "    \n",
    "def draw(img, size=(8, 8), cmap='jet'):\n",
    "    plt.figure(figsize=size)\n",
    "    plt.imshow(img, cmap=cmap)\n",
    "    plt.plot()\n",
    "    \n",
    "\n",
    "# from random import shuffle\n",
    "\n",
    "# def get_data():\n",
    "#     NUM = 0\n",
    "#     for img_path, seg_path in zip(glob.glob(IMG_ROOT + '/**/*T2.nii.gz'), glob.glob(IMG_ROOT + '/**/*Segmentation.nii.gz')):\n",
    "#         img = nib.load(img_path).get_data()\n",
    "#         seg = nib.load(seg_path).get_data()\n",
    "#         s = np.sum(seg.reshape((seg.shape[2], -1)).astype(np.int32), axis=1)\n",
    "#         indicies = s > 0\n",
    "#         LEN = indicies.shape[0]\n",
    "#         NUM += LEN\n",
    "#         indicies = indicies.reshape((-1))\n",
    "#         if NUM > LIMIT:\n",
    "#             raise StopIteration\n",
    "#         else:\n",
    "#             print(s)\n",
    "#             indicies = np.argwhere(indicies).reshape(-1)\n",
    "#             print(indicies)\n",
    "#             img2 = img[:, :, indicies]\n",
    "            \n",
    "#             seg2 = seg[:, :, indicies]\n",
    "            \n",
    "#             print(seg2.shape)\n",
    "#             print(np.sum(seg2.reshape((seg2.shape[2], -1)).astype(np.int32), axis=1))\n",
    "#             yield img2[:, :, 4], seg2[:, :, 4]\n",
    "        \n",
    "\n",
    "        \n",
    "def preprocess(data):\n",
    "    for data_ in tqdm(data, desc='Preprocessing'):\n",
    "        img = data_[0].astype(np.float32)\n",
    "        img = cv.medianBlur(img, 3)\n",
    "        img, _ = median_otsu(img, numpass=NUMPASS, median_radius=MEDIAN_RADIUS)\n",
    "        original_shape = img.shape\n",
    "        img = ((img - np.min(img)) / (np.max(img) - np.min(img))).astype(np.float32)\n",
    "        blurred = cv.blur(img, (19, 19))\n",
    "        edges = np.clip(img - blurred, 0.0, 1.0)\n",
    "        edges[edges > HIT] = 1.0\n",
    "        edges[edges <= HIT] = 0.0\n",
    "        edges = cv.dilate(edges, np.ones((2, 2)), iterations=1)\n",
    "        img = np.clip(img - edges, 0.0, 1.0)\n",
    "        yield (img, data_[1])\n",
    "    \n",
    "PATIENTS = [\n",
    "    ('104', 38),  \n",
    "    ('220', 25),\n",
    "    ('225', 35),\n",
    "    ('229', 14),\n",
    "    ('249', 35),\n",
    "    ('374', 45),\n",
    "    ('600', 25),\n",
    "    ('500', 33),\n",
    "    ('547', 39),\n",
    "    ('311', 39),\n",
    "    (357, 35),\n",
    "    (387, 16),\n",
    "    (345, 30),\n",
    "    (346, 48),\n",
    "    (351, 24),\n",
    "    (354, 28),\n",
    "    (355, 40),\n",
    "    (357, 36),\n",
    "    (359, 33),\n",
    "    (360, 24),\n",
    "    (361, 19),\n",
    "    (363, 23),\n",
    "    (365, 28),\n",
    "    (367, 28),\n",
    "    (371, 35),\n",
    "    (373, 34),\n",
    "    (374, 44),\n",
    "    (375, 43),\n",
    "    (377, 38),\n",
    "    (380, 37),\n",
    "    (383, 38),\n",
    "    (385, 36),\n",
    "    (388, 43),\n",
    "    (391, 25),\n",
    "    (394, 38),\n",
    "    (396, 37),\n",
    "    (492, 40),\n",
    "    (500, 34),\n",
    "    (506, 26),\n",
    "    (515, 25),\n",
    "    (516, 43),\n",
    "    (518, 49),\n",
    "    (519, 23),\n",
    "    (520, 33),\n",
    "] \n",
    "\n",
    "N = len(PATIENTS)\n",
    "print(N)\n",
    "\n",
    "DATA = list(get_data(IMG_ROOT, PATIENTS))\n",
    "PREP = list(preprocess(DATA))\n",
    "\n",
    "# INDEX = 0\n",
    "# orig = DATA[INDEX][0]\n",
    "# # image = PREP[INDEX][0]\n",
    "# gt = DATA[INDEX][1]\n",
    "\n",
    "# # res, labels = multivariate_kmeans(image)\n",
    "# # print('Dice (%)', calc_dice(res, gt))\n",
    "# # draw(orig)\n",
    "# # # draw(labels)\n",
    "# # # draw(res)\n",
    "# # draw(gt)\n",
    "\n",
    "# plt.figure(figsize=(8, 8))\n",
    "# plt.imshow(orig, 'gray')\n",
    "# plt.imshow(gt, alpha=0.3)\n",
    "\n",
    "# plt.plot()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# K Means"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def kmeans(img, n_clusters):\n",
    "    img = cv.blur(img, (15, 15))\n",
    "    x = img.reshape((-1, 1))\n",
    "    model = KMeans(n_clusters=n_clusters, random_state=0, algorithm='full', precompute_distances=False, n_init=1)\n",
    "    model.fit(x)\n",
    "    c_index = np.argmax(model.cluster_centers_.reshape((-1)))\n",
    "    flat = np.full(img.shape[0] * img.shape[1], 0, dtype=np.uint8)\n",
    "    flat[model.labels_ == c_index] = 1\n",
    "    mask = flat.reshape(img.shape)\n",
    "    mask = cv.dilate(mask, K1, iterations=1)\n",
    "    mask = cv.erode(mask, K1, iterations=1)\n",
    "    return mask"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Weighted Multivariate K Means"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def multivariate_kmeans(img, n_clusters):\n",
    "    f1 = cv.blur(img, (7, 7)).reshape((-1, 1))\n",
    "    f2 = cv.blur(img, (9, 9)).reshape((-1, 1)) \n",
    "    f3 = cv.blur(img, (5, 5)).reshape((-1, 1)) \n",
    "    f4 = img.reshape((-1, 1)) * 0.45\n",
    "    xc = np.linspace(0, 1.0, img.shape[1])\n",
    "    yc = np.linspace(0, 1.0, img.shape[0])\n",
    "    f5 = np.transpose([np.tile(xc, len(xc)), np.repeat(yc, len(yc))]).reshape((-1, 2)) * 0.015\n",
    "    f3 = cv.blur(img, (5, 5)).reshape((-1, 1)) \n",
    "    X = np.concatenate((f1, f2, f3, f4, f5), axis=1)\n",
    "    x = X\n",
    "    model = KMeans(n_clusters=n_clusters, random_state=0, algorithm='full', precompute_distances=False, n_init=1)\n",
    "    model.fit(x)\n",
    "    c_index = np.argmax(model.cluster_centers_[:,3])\n",
    "    flat = np.full(img.shape[0] * img.shape[1], 0, dtype=np.uint8)\n",
    "    flat[model.labels_ == c_index] = 1\n",
    "    mask = flat.reshape(img.shape)\n",
    "    mask = cv.erode(mask, K1, iterations=1)\n",
    "    mask = cv.dilate(mask, K1, iterations=1)\n",
    "    return mask"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Fuzzy C Means"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def fcm(img, n_clusters):\n",
    "    M = 2\n",
    "    img = cv.blur(img, (15, 15))\n",
    "    flat = img.reshape((1, -1))\n",
    "    c, u, a1, a2, a3, a4, a5 = cmeans(flat, n_clusters, M, EPS, 50)\n",
    "    tumor_index = np.argmax(c, axis=0)\n",
    "    defuz = np.argmax(u, axis=0)\n",
    "    mask = np.full(defuz.shape[0], 0, dtype=np.uint16)\n",
    "    mask[defuz == tumor_index] = 1\n",
    "    mask = mask.reshape(img.shape)\n",
    "    mask = cv.erode(mask, K1, iterations=1)\n",
    "    mask = cv.dilate(mask, K1, iterations=1)\n",
    "    return mask"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Gaussian Mixture"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def gaussian_mixture(img, n_components):\n",
    "    img = cv.blur(img, (15, 15))\n",
    "    x = img.reshape((-1, 1))\n",
    "    model = GaussianMixture(n_components=n_components, covariance_type='spherical')\n",
    "    model.fit(x)\n",
    "    labels = model.predict(x)\n",
    "    c_index = np.argmax(model.means_)\n",
    "    flat = np.full(img.shape[0] * img.shape[1], 0, dtype=np.uint8)\n",
    "    flat[labels == c_index] = 1\n",
    "    mask = flat.reshape(img.shape)\n",
    "    mask = cv.erode(mask, K1, iterations=1)\n",
    "    mask = cv.dilate(mask, K1, iterations=1)\n",
    "    return mask"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Binary thresholding"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def find_threshold_cpu(img, max_it=100, eps=0.001):\n",
    "    cols = img.shape[1]\n",
    "\n",
    "    mid = cols // 2\n",
    "    p1 = img[:, :mid].reshape((-1))\n",
    "    p2 = img[:, mid:].reshape((-1))\n",
    "    img = img.reshape((-1))\n",
    "\n",
    "    it = -1\n",
    "    while True:\n",
    "        t1 = np.mean(p1)\n",
    "        t2 = np.mean(p2)\n",
    "\n",
    "        t = (t1 + t2) / 2.0\n",
    "        it += 1\n",
    "        if it >= max_it or abs(t1 - t2) <= eps:\n",
    "            return t\n",
    "        else:\n",
    "            p1 = img[img < t]\n",
    "            p2 = img[img >= t]\n",
    "        \n",
    "def threshold(img):\n",
    "    t = find_threshold_cpu(img, max_it=MAX_IT, eps=EPS)\n",
    "    mask = (img > t).astype(np.int16)\n",
    "    mask = cv.erode(mask, K1, iterations=1)\n",
    "    mask = cv.dilate(mask, K1, iterations=1)\n",
    "    return mask"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pycuda.driver as cuda\n",
    "import pycuda.autoinit\n",
    "from pycuda.compiler import SourceModule\n",
    "\n",
    "import cv2 as cv\n",
    "import numpy as np\n",
    "from dipy.segment.mask import median_otsu\n",
    "from pydicom import Dataset\n",
    "\n",
    "from jinja2 import Template\n",
    "import numpy as np\n",
    "\n",
    "BLOCKDIM = 1024\n",
    "\n",
    "SRC = '''\n",
    "    #define N {{N}}\n",
    "    #define BLOCKDIM 1024\n",
    "    \n",
    "    struct Cluster{\n",
    "        float sum;\n",
    "        int count;\n",
    "    };\n",
    "    \n",
    "    __device__ Cluster clusters_d[(N + BLOCKDIM - 1) / BLOCKDIM];\n",
    "    \n",
    "    __device__ float euclidian_dist(const float a, const float b){\n",
    "        float dist = a - b;\n",
    "        return hypotf(dist, dist);\n",
    "    }\n",
    "    \n",
    "    __global__ void relabel(const float* src, const float* clusters, int n, int nClusters, int* labels){\n",
    "        int pos = threadIdx.x + blockIdx.x * blockDim.x;\n",
    "        if(pos < n){\n",
    "            float minDist = 1.0f;\n",
    "            int clusterIndex = 0;\n",
    "            for(int c = 0; c < nClusters; c++){\n",
    "                float dist = euclidian_dist(src[pos], clusters[c]);\n",
    "                if(dist <= minDist){\n",
    "                    clusterIndex = c;\n",
    "                    minDist = dist;\n",
    "                }\n",
    "            }\n",
    "            labels[pos] = clusterIndex;\n",
    "        }\n",
    "    }\n",
    "    \n",
    "    __global__ void calculateClusters(const float* src, const int* labels, int n, int clusterIndex){\n",
    "        extern __shared__ Cluster _clusters[];\n",
    "        int pos = threadIdx.x + blockIdx.x * blockDim.x;\n",
    "        int tid = threadIdx.x;\n",
    "        _clusters[tid] = Cluster();\n",
    "        _clusters[tid].sum = 0.0f;\n",
    "        _clusters[tid].count = 0;\n",
    "        if(pos < n && labels[pos] == clusterIndex){\n",
    "            _clusters[tid].sum = src[pos];\n",
    "            _clusters[tid].count = 1;\n",
    "        }\n",
    "        __syncthreads();\n",
    "        for(unsigned int stride = blockDim.x / 2; stride > 0; stride /= 2){\n",
    "            if(threadIdx.x < stride){\n",
    "                _clusters[tid].sum += _clusters[tid + stride].sum;\n",
    "            }\n",
    "            __syncthreads();\n",
    "            if(threadIdx.x < stride){\n",
    "                _clusters[tid].count += _clusters[tid + stride].count;\n",
    "            }\n",
    "            __syncthreads();\n",
    "        }\n",
    "        __syncthreads();\n",
    "        if(threadIdx.x == 0){\n",
    "            clusters_d[blockIdx.x].sum = _clusters[0].sum;\n",
    "            clusters_d[blockIdx.x].count = _clusters[0].count;\n",
    "        }\n",
    "    }\n",
    "    \n",
    "    __global__ void findCenters(int n, int clusterIndex, float* dst){\n",
    "        extern __shared__ Cluster _clusters[];\n",
    "        int pos = threadIdx.x + blockIdx.x * blockDim.x;\n",
    "        int tid = threadIdx.x;\n",
    "        _clusters[tid] = clusters_d[pos];\n",
    "        __syncthreads();\n",
    "        for(unsigned int stride = blockDim.x / 2; stride > 0; stride /= 2){\n",
    "            if(tid < stride){\n",
    "                _clusters[tid].sum += _clusters[tid + stride].sum;\n",
    "            }\n",
    "            __syncthreads();\n",
    "            if(tid < stride){\n",
    "                _clusters[tid].count += _clusters[tid + stride].count;\n",
    "            }\n",
    "            __syncthreads();\n",
    "        }\n",
    "        __syncthreads();\n",
    "        if(tid == 0){\n",
    "            dst[clusterIndex] = _clusters[0].count > 0 ? _clusters[0].sum / (_clusters[0].count * 1.0f) : 0.0f;\n",
    "        }\n",
    "    }\n",
    "'''\n",
    "\n",
    "def cuda_kmeans(img, n_clusters):\n",
    "        w = img.shape[1]\n",
    "        h = img.shape[0]\n",
    "        n = w * h\n",
    "        max_it = 30\n",
    "\n",
    "        img = cv.blur(img, (13, 13))\n",
    "        src = img.reshape((-1))\n",
    "\n",
    "        module = SourceModule(Template(SRC).render(N=n))\n",
    "        relabel = module.get_function('relabel')\n",
    "        calculate_clusters = module.get_function('calculateClusters')\n",
    "        find_centers = module.get_function('findCenters')\n",
    "\n",
    "        t0 = time() * 1000\n",
    "        centers = np.random.rand(n_clusters).astype(np.float32)\n",
    "\n",
    "        # Image\n",
    "        src_gpu = cuda.mem_alloc(src.nbytes)\n",
    "        cuda.memcpy_htod(src_gpu, src)\n",
    "\n",
    "        # Cluster centers\n",
    "\n",
    "        centers_gpu = cuda.mem_alloc(centers.nbytes)\n",
    "        cuda.memcpy_htod(centers_gpu, centers)\n",
    "\n",
    "        # Labels\n",
    "        labels = np.empty_like(src).astype(np.int32)\n",
    "        labels_gpu = cuda.mem_alloc(labels.nbytes)\n",
    "        cuda.memcpy_htod(labels_gpu, labels)\n",
    "        \n",
    "        start = cuda.Event()\n",
    "        end = cuda.Event()\n",
    "        \n",
    "        t1 = time() * 1000\n",
    "        \n",
    "        start.record()\n",
    "        for it in range(max_it):\n",
    "            relabel(src_gpu, centers_gpu, np.int32(n), np.int32(n_clusters), labels_gpu,\n",
    "                    block=(BLOCKDIM, 1, 1), grid=((n + BLOCKDIM - 1) // BLOCKDIM, 1))\n",
    "            for c in range(n_clusters):\n",
    "                calculate_clusters(src_gpu, labels_gpu, np.int32(n), np.int32(c),\n",
    "                                   block=(BLOCKDIM, 1, 1), grid=((n + BLOCKDIM - 1) // BLOCKDIM, 1),\n",
    "                                   shared=8 * BLOCKDIM)\n",
    "                find_centers(np.int32(n), np.int32(c), centers_gpu,\n",
    "                             block=((n + BLOCKDIM - 1) // BLOCKDIM, 1, 1), grid=((1, 1)),\n",
    "                             shared=8 * (n + BLOCKDIM - 1) // BLOCKDIM)\n",
    "\n",
    "        \n",
    "        end.record()\n",
    "        end.synchronize()\n",
    "        msecs = end.time_since(start)*1e-3\n",
    "        msecs *= 1000\n",
    "        msecs += t1 - t0\n",
    "        \n",
    "        t0 = time() * 1000\n",
    "        cuda.memcpy_dtoh(labels, labels_gpu)\n",
    "        cuda.memcpy_dtoh(centers, centers_gpu)\n",
    "#\n",
    "        labels = labels.reshape((-1))\n",
    "        c_index = np.argmax(centers)\n",
    "        flat = np.full(n, 0, dtype=np.uint8)\n",
    "        flat[labels == c_index] = 1\n",
    "        mask = flat.reshape((h, w))\n",
    "        mask = cv.erode(mask, kernel=K1, iterations=1)\n",
    "        mask = cv.dilate(mask, kernel=K1, iterations=1)\n",
    "        t1 = time() * 1000\n",
    "        msecs += t1 - t0\n",
    "        return mask, msecs\n",
    "\n",
    "cuda_kmeans.is_cuda = True"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Profile average time and accuracy (Dice-Sørensen) per image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "import collections\n",
    "ALGS = collections.OrderedDict()\n",
    "ALGS['K Means'] = kmeans\n",
    "# ALGS['Fuzzy C Means'] = fcm\n",
    "ALGS['Gaussian Mixture'] = gaussian_mixture\n",
    "# ALGS['Multivariate K Means'] = multivariate_kmeans\n",
    "ALGS['CUDA K Means'] = cuda_kmeans\n",
    "\n",
    "\n",
    "def t(alpha, gl):\n",
    "    return scipy.stats.t.ppf(1-(alpha/2), gl)\n",
    "\n",
    "TP_INDEX = 0\n",
    "FP_INDEX = 1\n",
    "FN_INDEX = 2\n",
    "DC_INDEX = 3\n",
    "PR_INDEX = 4\n",
    "RC_INDEX = 5\n",
    "F1_INDEX = 6\n",
    "\n",
    "def calc_accuracy(seg, gt):\n",
    "    size = seg.shape[0] * seg.shape[1]\n",
    "    tp = np.sum(seg[gt==1])\n",
    "    tn = np.sum(seg[gt==0])\n",
    "    \n",
    "    fp = np.sum((gt==0) & (seg==1))\n",
    "    fn = np.sum((gt==1) & (seg==0))\n",
    "    dc = 2.0 * tp / (np.sum(gt) + np.sum(seg)) * 100.0\n",
    "\n",
    "    precision = tp / (tp + fp) * 100.0 \n",
    "    recall = tp / (tp + fn) * 100.0\n",
    "\n",
    "    tp = tp * 100.0 / size\n",
    "    fp = fp * 100.0 / size\n",
    "    fn = fn * 100.0 / size\n",
    "    return dc\n",
    "\n",
    "def profile(alg, data, n_clusters):\n",
    "    avg_time = []\n",
    "    avg_accuracy = []\n",
    "    for data_item in iter(data):\n",
    "        time_it = []\n",
    "        accuracy_it = []\n",
    "        img, gt = data_item[0], data_item[1]\n",
    "        for i in range(N_ITERS):\n",
    "            duration = 0.0\n",
    "            if getattr(alg, 'is_cuda', False):\n",
    "                seg, duration = alg(img, n_clusters)\n",
    "            else:\n",
    "                t0 = time()\n",
    "                seg = alg(img, n_clusters)\n",
    "                t1 = time()\n",
    "                duration = (t1 - t0) * 1000.0\n",
    "            \n",
    "            time_it.append([duration])\n",
    "            accuracy_it.append(calc_accuracy(seg, gt))\n",
    "        avg_time.append(np.mean(time_it))\n",
    "        avg_accuracy.append(np.max(accuracy_it))\n",
    "    return avg_time, avg_accuracy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e9c57ac9607c4906bafb9c180e37602c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Collecting data', max=7), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "================================\n",
      "N CLUSTERS =  3\n",
      "================================\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "799bf4f69d404f66aeb054e4a1dcedaa",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=3), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average time per image\n",
      "{'K Means': [24.35152530670166, 41.110825538635254, 93.92101764678955, 15.60981273651123, 20.098114013671875, 22.975778579711914, 20.09742259979248, 29.53495979309082, 46.048712730407715, 24.935173988342285, 17.861199378967285, 135.3639841079712, 18.546438217163086, 16.664671897888184, 19.30069923400879, 17.43783950805664, 22.469401359558105, 13.855504989624023, 34.52329635620117, 29.20403480529785, 85.55412292480469, 19.8972225189209, 22.304511070251465, 18.211913108825684, 25.382351875305176, 26.38227939605713, 24.300003051757812, 15.504002571105957, 22.51424789428711, 14.457082748413086, 19.70043182373047, 25.445961952209473, 39.24908638000488, 14.299154281616211, 18.61262321472168, 17.231225967407227, 40.77267646789551, 50.51264762878418, 18.213915824890137, 78.3273458480835, 33.90040397644043, 61.98720932006836, 31.06672763824463, 16.4337158203125], 'Gaussian Mixture': [190.4585599899292, 208.15799236297607, 277.40352153778076, 222.4971055984497, 231.5296173095703, 170.6127643585205, 211.86141967773438, 367.5898551940918, 236.42375469207764, 189.89944458007812, 222.63669967651367, 991.1643981933594, 198.06363582611084, 204.6999454498291, 211.30659580230713, 194.8375701904297, 197.9773759841919, 200.97081661224365, 246.0216522216797, 220.73657512664795, 269.1112756729126, 200.47802925109863, 197.65393733978271, 206.81684017181396, 189.294171333313, 204.94837760925293, 202.53632068634033, 201.51894092559814, 186.59992218017578, 222.15380668640137, 284.58075523376465, 234.60192680358887, 259.4388961791992, 249.27325248718262, 195.05116939544678, 152.43792533874512, 251.39575004577637, 257.1709632873535, 183.62700939178467, 882.2826147079468, 171.91526889801025, 160.67073345184326, 257.8411102294922, 220.4460859298706], 'CUDA K Means': [36.83480854034424, 22.331812477111818, 22.06997604370117, 22.561250686645508, 23.39396114349365, 21.645863342285157, 21.334130859375, 21.274946022033692, 20.765373611450194, 21.759821319580077, 20.602778816223143, 70.45437240600586, 17.286894607543946, 17.218372535705566, 17.320579719543456, 17.251523399353026, 17.34675178527832, 17.437932205200195, 17.424168968200682, 17.27665195465088, 17.430776023864745, 17.95826950073242, 17.4803897857666, 17.38807201385498, 17.37903594970703, 17.31243133544922, 17.42189826965332, 17.394322013854982, 17.358981323242187, 17.345237922668456, 17.24531078338623, 17.247864151000975, 17.187650299072267, 17.251095962524413, 17.42890567779541, 18.046592903137206, 17.549790573120116, 17.738489151000977, 17.468259239196776, 61.61156044006348, 17.018162536621094, 17.268718242645264, 17.336749172210695, 17.372235679626463]}\n",
      "\n",
      "\n",
      "Average accuracy per image\n",
      "{'K Means': [85.30834340991535, 10.692481161652275, 31.574833919499806, 86.17295168862638, 81.11145800811738, 93.35907335907336, 92.52183406113537, 15.658273831892771, 80.59467918622848, 21.53028692879915, 88.56447688564477, 40.09177775841538, 88.99467936152338, 70.41123370110331, 88.5451197053407, 93.52170916609235, 80.65984072810012, 85.93350383631714, 90.51825677267374, 81.71641791044776, 30.011933174224342, 92.73356401384083, 77.54021584198738, 87.35769922109047, 10.416666666666668, 72.36045729657027, 92.27306826706678, 88.29758349171871, 95.54940910233844, 91.19266055045871, 25.15267570644861, 87.22576629842851, 8.31291405753698, 81.93947282785552, 85.37398761314911, 3.8819095477386933, 18.851201636270666, 14.114696157792725, 95.04264392324095, 95.9902399349329, 73.79335001787629, 33.93025447690858, 80.05930318754633, 86.71742808798646], 'Gaussian Mixture': [75.20397304008513, 56.42540620384048, 8.948399738732855, 61.1764705882353, 64.10767696909272, 94.84276729559748, 92.15497035709362, 0.12705583398037693, 68.08396124865446, 56.913580246913575, 82.62859234852645, 38.963712790597924, 73.89199255121042, 87.26450640542576, 78.89833940866747, 91.54497926807284, 69.9228791773779, 78.73373470576811, 65.12, 68.49700809574094, 8.236567604802204, 75.83289264939185, 64.85647788983708, 79.12324829320877, 40.38972542072631, 65.72734196496572, 92.3076923076923, 85.7500782962731, 91.85714285714286, 81.08747044917257, 22.69543289436817, 60.528128587830075, 64.60176991150442, 63.55889724310777, 83.90941597139452, 3.931293093020443, 0.017691287041132243, 0.26961827728111254, 94.19063733784546, 91.2210935811541, 85.98912588874947, 28.026469443363176, 55.28281750266809, 66.96517412935323], 'CUDA K Means': [85.09732360097324, 89.27628103539355, 47.54491017964072, 86.55077767612077, 83.55326681686515, 94.866693195384, 92.8248044999314, 15.929654645087293, 78.0980906027705, 82.7024436990896, 88.46755957557836, 39.71483220309511, 87.80622075419762, 71.30523108176739, 87.5558867362146, 93.62957819822948, 81.21000581733566, 86.12073682290716, 89.96372430471584, 82.11920529801324, 35.96077006901562, 92.35528060097215, 76.55662348516506, 87.2198667474258, 72.98198799199467, 72.3679727427598, 94.03616775682954, 88.57693366363887, 96.21346886912325, 93.2387706855792, 58.13658288905814, 88.24650571791614, 10.220932827493863, 83.89463154384795, 86.80154142581888, 3.8507519033536775, 20.28985507246377, 13.740288568257492, 95.44962080173349, 95.995115995116, 77.76940467219292, 35.749751737835155, 80.57663125948406, 85.92150170648463]}\n",
      "\n",
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b745b5e1a41a4ad0a038800d3f804e3e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Avg time per algorithm (N_CLUSTERS = 3)', max=3), HTML(value=…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "0cb20d3c04dc44e780a1aee6165194c3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Avg acc per algorithm (N_CLUSTERS = 3)', max=3), HTML(value='…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "================================\n",
      "N CLUSTERS =  4\n",
      "================================\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d4d174b9aaa749fe906567038f4a48e3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=3), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average time per image\n",
      "{'K Means': [34.55379009246826, 52.33147144317627, 40.482401847839355, 25.377368927001953, 27.9252290725708, 58.33179950714111, 92.64640808105469, 42.62828826904297, 31.56440258026123, 33.90381336212158, 34.4771146774292, 116.2445068359375, 29.966187477111816, 27.884507179260254, 29.745769500732422, 36.858558654785156, 34.833478927612305, 31.784582138061523, 52.022504806518555, 60.6548547744751, 72.03452587127686, 35.86456775665283, 66.94731712341309, 23.981499671936035, 24.672245979309082, 201.41758918762207, 87.2539758682251, 34.247660636901855, 30.23066520690918, 24.625706672668457, 50.49629211425781, 61.48691177368164, 30.083775520324707, 62.25638389587402, 52.9735803604126, 42.10395812988281, 31.957435607910156, 28.225064277648926, 26.433110237121582, 125.42927265167236, 61.03827953338623, 25.53117275238037, 38.55853080749512, 24.89769458770752], 'Gaussian Mixture': [211.57546043395996, 224.04921054840088, 252.81598567962646, 238.27857971191406, 214.07809257507324, 244.58379745483398, 291.32094383239746, 305.39960861206055, 246.01593017578125, 208.0822229385376, 241.67828559875488, 1099.1567850112915, 234.76955890655518, 240.13681411743164, 225.0211477279663, 258.3622694015503, 218.5788631439209, 243.33181381225586, 228.0766487121582, 228.92279624938965, 354.9196720123291, 238.88282775878906, 231.26885890960693, 243.49195957183838, 218.72937679290771, 266.6191101074219, 251.05204582214355, 201.0838508605957, 288.39385509490967, 228.80520820617676, 225.7441520690918, 213.13002109527588, 222.85125255584717, 216.72213077545166, 198.0116367340088, 218.46251487731934, 236.67728900909424, 287.8535985946655, 230.72922229766846, 1076.4047861099243, 217.83154010772705, 186.36860847473145, 220.84119319915771, 228.06158065795898], 'CUDA K Means': [38.16726226806641, 26.36925849914551, 24.865032386779784, 24.050452041625977, 23.450982093811035, 22.915615463256835, 23.085965156555176, 23.557848167419433, 23.559387588500975, 23.331110382080077, 23.385511779785155, 80.71144104003906, 21.379326820373535, 21.340747833251953, 21.36584587097168, 21.378608894348144, 21.317017555236816, 21.39227466583252, 21.36177616119385, 21.356786155700682, 21.427425384521484, 21.44029483795166, 21.36987724304199, 21.55924892425537, 21.433473205566408, 21.26846561431885, 21.37935028076172, 21.32751636505127, 21.311457633972168, 21.326778411865234, 21.404199409484864, 21.40217399597168, 21.395039367675782, 21.3745174407959, 21.494205665588378, 21.53818702697754, 21.30662097930908, 21.451762771606447, 21.37277545928955, 82.4749969482422, 21.615534782409668, 23.286964225769044, 21.714706039428712, 22.90884609222412]}\n",
      "\n",
      "\n",
      "Average accuracy per image\n",
      "{'K Means': [84.22674060382008, 88.52119958634952, 77.19107240065324, 85.95117457392907, 83.54430379746836, 95.17241379310344, 82.85536159600998, 74.31984074319841, 79.91202346041055, 81.17529880478087, 73.53057199211045, 45.32381508048186, 89.64131994261119, 74.73118279569893, 87.70491803278688, 83.57572231917781, 81.85554903112156, 84.7970479704797, 89.8444647758463, 83.64703006599854, 49.74647887323943, 92.19232465813852, 66.71736375158429, 86.10337356855463, 71.83288409703505, 71.54639175257732, 94.69964664310953, 88.53096362121633, 84.76702508960572, 92.8605654048874, 80.93951361463313, 88.55656697009103, 75.08196721311475, 85.60338743824983, 86.28211342704799, 7.471432757105186, 39.11649447809049, 87.31942215088283, 95.05004057343793, 95.91220626631853, 87.34069668649109, 86.38036809815951, 80.0, 85.43273350471294], 'Gaussian Mixture': [84.91905708605509, 86.48648648648648, 78.42484747642817, 85.36850693500136, 77.8140778140778, 86.32840028188865, 92.60318757662444, 83.72093023255815, 81.67865707434053, 81.96411709159584, 91.47405479871422, 56.931309109009455, 89.29474571508851, 67.36942980354576, 89.62843295638126, 93.30379962514908, 73.30316742081448, 90.57574782888389, 89.383272317657, 82.054890921886, 63.0912925579994, 92.17002237136465, 73.04630969609262, 85.2530779753762, 70.625, 71.36563876651982, 84.29919137466307, 85.50688360450563, 77.21196690006366, 89.53749438706781, 81.94570135746606, 88.03676121058469, 79.3073593073593, 77.76777677767777, 78.29751369574379, 11.134211134211135, 60.313771063335274, 86.06060606060606, 95.09883563498511, 94.6087621962885, 80.03101977510663, 90.47619047619048, 78.18052594171996, 80.56968463886062], 'CUDA K Means': [84.50355706773894, 88.68435300487276, 74.97403946002078, 85.87291795188156, 84.54154253558424, 95.265780730897, 85.3922315308454, 79.68069666182875, 83.30676383511728, 82.06656731246895, 87.70577933450087, 55.3979376744716, 89.85340615119287, 77.67207610520425, 86.77966101694915, 93.36609336609337, 82.22490931076179, 85.26277104005881, 89.1432106402722, 83.34965719882467, 55.0328022492971, 92.14894571556752, 75.91209729037764, 85.38048626460373, 73.08743169398907, 71.56282631395754, 95.44364508393285, 88.88888888888889, 88.9144831557731, 93.89763779527559, 79.13300492610837, 89.15860259143841, 79.51388888888889, 85.25052928722654, 87.88927335640139, 7.755162680136681, 49.57825679475164, 89.14664457332229, 95.42089388538525, 95.92069837643795, 89.84615384615384, 83.86336866902238, 81.60642570281125, 84.84065460809647]}\n",
      "\n",
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7859af3e779e47348147a33a39dc1be0",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Avg time per algorithm (N_CLUSTERS = 4)', max=3), HTML(value=…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "919995b828374e7bad0d259173a4f2bb",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Avg acc per algorithm (N_CLUSTERS = 4)', max=3), HTML(value='…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "================================\n",
      "N CLUSTERS =  5\n",
      "================================\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b607c49b3c6242d2a98e6f592b082d41",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=3), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average time per image\n",
      "{'K Means': [68.91276836395264, 39.756155014038086, 57.72838592529297, 33.25331211090088, 36.83369159698486, 34.5217227935791, 110.04400253295898, 75.43168067932129, 52.512383460998535, 32.140493392944336, 41.103339195251465, 246.54123783111572, 50.45497417449951, 42.668819427490234, 34.29560661315918, 48.590779304504395, 61.9152307510376, 32.48310089111328, 32.11197853088379, 38.359999656677246, 105.74750900268555, 56.99045658111572, 56.77385330200195, 26.358580589294434, 40.527963638305664, 69.10595893859863, 36.74812316894531, 64.14153575897217, 40.56074619293213, 48.123764991760254, 29.063796997070312, 25.054407119750977, 69.67425346374512, 42.96398162841797, 25.524163246154785, 204.88147735595703, 47.88665771484375, 73.47919940948486, 98.55732917785645, 150.52828788757324, 77.72603034973145, 83.54120254516602, 48.20237159729004, 37.80224323272705], 'Gaussian Mixture': [278.5569667816162, 331.57217502593994, 300.95906257629395, 269.95980739593506, 276.54125690460205, 266.0139322280884, 420.4505920410156, 318.5741662979126, 252.38728523254395, 251.07266902923584, 267.1112537384033, 1124.6596097946167, 254.1152000427246, 304.452919960022, 295.466685295105, 279.6405076980591, 260.2187395095825, 249.54149723052979, 265.6853675842285, 271.2389945983887, 329.83808517456055, 243.4896469116211, 287.63771057128906, 261.55340671539307, 236.63806915283203, 327.6419401168823, 311.36977672576904, 239.21477794647217, 271.71971797943115, 252.78608798980713, 308.90393257141113, 266.34228229522705, 260.04114151000977, 284.6161127090454, 240.77208042144775, 276.14145278930664, 277.6501417160034, 275.78022480010986, 252.90727615356445, 1208.347725868225, 279.4468641281128, 214.00165557861328, 279.5971155166626, 288.2431745529175], 'CUDA K Means': [51.36226348876953, 31.588138580322266, 29.75700435638428, 29.30229206085205, 27.96253776550293, 26.782761001586913, 26.231597709655762, 27.46450710296631, 26.46287593841553, 26.42000503540039, 27.28367462158203, 99.35952072143554, 27.30888080596924, 27.553972434997558, 26.267441177368163, 26.951872634887696, 28.002614974975586, 26.397177505493165, 28.137968444824217, 26.29880027770996, 26.59925994873047, 26.40015983581543, 26.454546165466308, 26.379789924621583, 26.26603126525879, 27.94705581665039, 26.89466075897217, 27.818528175354004, 26.892314147949218, 27.139554595947267, 27.6415678024292, 26.352092742919922, 27.216473388671876, 28.484056854248045, 28.185053443908693, 28.06305503845215, 27.235783195495607, 26.463237380981447, 26.503877258300783, 99.38797531127929, 27.164243125915526, 28.02345542907715, 26.372493171691893, 27.898932647705077]}\n",
      "\n",
      "\n",
      "Average accuracy per image\n",
      "{'K Means': [80.39735099337749, 82.85209192692987, 73.64341085271317, 71.4913051115405, 85.12669849430775, 94.84708839547548, 65.52094522019334, 80.79268292682927, 77.05041384499623, 80.6841046277666, 72.67672756155679, 50.646410050398075, 89.61479198767334, 81.45968216597998, 69.54624781849913, 82.5508607198748, 82.31046931407943, 69.34505126595522, 84.84648398811488, 84.8852545502506, 61.74193548387097, 88.62671071260029, 59.4048594048594, 79.58600999286224, 70.11331444759207, 56.694286960314, 87.5776397515528, 83.88349514563107, 82.5928180158247, 90.55201698513801, 81.20789779326365, 86.57148020965118, 83.42440801457195, 84.82897384305835, 88.04071246819338, 8.088354718206919, 70.7492795389049, 86.80851063829788, 92.74356750505926, 93.86466924760141, 92.14515388148828, 90.09247027741083, 81.70829975825947, 78.12215132178669], 'Gaussian Mixture': [84.41717791411043, 89.35303107488538, 79.7945205479452, 85.23790158601058, 84.22097254383063, 87.21965112139551, 64.28184281842817, 80.12093726379442, 77.173640691556, 81.34500230308613, 67.38366080661841, 44.763132452465356, 89.60093212933295, 68.82352941176471, 89.72776769509981, 81.78775349478244, 75.625, 63.905841325196164, 90.58374889997066, 83.28899637243047, 60.309973045822105, 92.46861924686193, 66.29299363057325, 85.20645337708504, 71.1168164313222, 57.53779697624191, 90.5644118914261, 85.13470681458003, 76.85924034471753, 92.84351145038168, 82.27988467509427, 88.74807330022264, 70.51578137028484, 83.57843137254902, 84.36578171091446, 11.577424023154848, 71.73119065010957, 87.5, 94.89822891884747, 96.16895086960007, 92.50234301780694, 88.63049095607235, 79.41818181818182, 86.573485811097], 'CUDA K Means': [81.19461765671151, 83.11688311688312, 76.12412919569348, 85.68333075375328, 85.71428571428571, 81.35593220338984, 80.78082843993018, 83.53808353808354, 77.41935483870968, 81.5670517327976, 73.40572556762093, 49.14576517629953, 89.97493734335839, 89.71428571428571, 79.28946306015341, 83.93582060699786, 82.71604938271605, 68.79596553897878, 85.33945555919973, 84.13942177789787, 61.61490683229813, 88.47795163584638, 71.78403755868544, 83.8070059484468, 72.57731958762886, 66.94152923538232, 94.68954248366013, 83.93613554903877, 82.91350531107739, 93.26530612244898, 81.15942028985508, 89.27622841965471, 86.52751423149905, 85.33333333333334, 88.9002557544757, 12.179487179487179, 74.50331125827815, 88.90742285237698, 93.25648414985591, 94.14628005539004, 92.22011385199241, 89.83957219251337, 82.59041211101767, 82.20264317180617]}\n",
      "\n",
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f09629a6c5c04fec971c7e3d0dcbfabf",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Avg time per algorithm (N_CLUSTERS = 5)', max=3), HTML(value=…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "718ea438f8f8424e820020be6272cc18",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Avg acc per algorithm (N_CLUSTERS = 5)', max=3), HTML(value='…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "================================\n",
      "N CLUSTERS =  6\n",
      "================================\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2b32486d28b9444990f4d3c34faf5a1f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=3), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average time per image\n",
      "{'K Means': [61.71731948852539, 62.410831451416016, 97.3686933517456, 48.74606132507324, 44.50550079345703, 72.84531593322754, 128.0198335647583, 89.11230564117432, 43.166184425354004, 63.259100914001465, 53.93989086151123, 215.71178436279297, 98.04544448852539, 85.47756671905518, 150.74551105499268, 100.51696300506592, 82.81059265136719, 44.35310363769531, 50.13110637664795, 51.450347900390625, 56.996726989746094, 55.286598205566406, 133.21335315704346, 52.8125524520874, 56.75954818725586, 47.241830825805664, 129.69329357147217, 72.72536754608154, 47.825050354003906, 89.73920345306396, 39.59000110626221, 38.730740547180176, 131.91115856170654, 70.99182605743408, 66.81561470031738, 87.85502910614014, 78.67660522460938, 100.3673791885376, 104.69973087310791, 235.18743515014648, 84.31072235107422, 91.15135669708252, 48.06082248687744, 116.43178462982178], 'Gaussian Mixture': [277.14829444885254, 314.853572845459, 339.54484462738037, 342.8267002105713, 340.39599895477295, 325.86236000061035, 383.62717628479004, 368.9663887023926, 307.72974491119385, 302.1206855773926, 357.1157217025757, 1251.5023946762085, 348.34611415863037, 329.8680067062378, 332.7256202697754, 334.08727645874023, 283.054518699646, 298.7297058105469, 299.4272232055664, 312.3791217803955, 338.4124517440796, 311.77706718444824, 330.5973768234253, 311.039137840271, 322.871470451355, 350.8226156234741, 375.0112533569336, 347.91393280029297, 316.4630651473999, 345.38283348083496, 307.6881170272827, 312.4369144439697, 324.1575002670288, 379.9147605895996, 344.00291442871094, 328.2978296279907, 340.02811908721924, 307.8652858734131, 347.37887382507324, 1308.7818384170532, 303.45520973205566, 263.6158227920532, 329.0735960006714, 286.1727714538574], 'CUDA K Means': [52.97322006225586, 37.02572746276856, 33.78124732971192, 31.638967514038086, 32.12889347076416, 33.20379009246826, 31.866510772705077, 31.446688652038574, 31.236613845825197, 31.220482444763185, 31.233965110778808, 114.7094825744629, 31.16709156036377, 31.187290000915528, 31.93098850250244, 31.34336929321289, 31.23945560455322, 32.14265041351318, 31.93186721801758, 31.413067245483397, 31.430831336975096, 33.50787544250488, 33.85258541107178, 31.483473587036134, 31.483926391601564, 31.275463676452638, 31.4591007232666, 32.76648082733154, 31.237350082397462, 31.623837089538576, 31.991589164733888, 32.25840873718262, 32.02317161560059, 31.266205406188966, 32.429592514038085, 32.5052770614624, 32.82024574279785, 32.25091342926025, 32.53668727874756, 116.77869720458985, 32.647911643981935, 31.378646850585938, 32.459038162231444, 31.369648742675782]}\n",
      "\n",
      "\n",
      "Average accuracy per image\n",
      "{'K Means': [79.90669776741086, 81.0843373493976, 70.47872340425532, 70.9700176366843, 85.0481125092524, 77.25450901803607, 63.492640377975654, 83.51648351648352, 75.42566709021601, 72.3568281938326, 72.01596806387226, 26.774256270659148, 85.61939554965127, 88.01775147928994, 66.39857015192135, 81.32041905514924, 75.27559055118111, 67.34650520501381, 84.28761651131825, 84.72560151392268, 58.50104094378904, 70.06087437742114, 58.53793103448276, 78.35875090777051, 69.39068100358423, 54.20981539846916, 78.57829010566762, 82.65440210249672, 80.44623489308955, 88.70703764320785, 80.50847457627118, 85.71952337305224, 85.95825426944972, 78.62728062554301, 82.21680876979293, 15.726495726495726, 75.04187604690117, 86.79896462467644, 79.109364767518, 93.77817094233119, 89.70736629667003, 87.03703703703704, 81.11201962387572, 75.33799533799534], 'Gaussian Mixture': [84.52599388379205, 88.78600823045268, 76.08286252354048, 80.18312622629169, 84.27672955974843, 74.55197132616487, 64.4043321299639, 80.12093726379442, 77.90784557907845, 79.71237801746275, 66.27737226277372, 24.343152866242036, 89.8041474654378, 87.8721859114016, 90.17341040462428, 81.17949732831981, 75.37372147915028, 63.28005255090869, 90.56271981242674, 83.97135095085207, 60.71547977265128, 91.9454770755886, 59.6395412342982, 79.6718972895863, 71.13665389527458, 56.049274087109545, 82.74582560296847, 84.76736775015934, 75.82877373672353, 93.34637964774952, 82.18198077353007, 88.3989501312336, 66.86131386861314, 79.94869602394186, 84.37869822485207, 16.133518776077885, 75.77019150707743, 87.34491315136476, 80.90322580645162, 84.12376830279031, 89.96990972918756, 90.47619047619048, 80.39664378337147, 87.58389261744966], 'CUDA K Means': [80.55739880557398, 80.8227465214761, 73.01173402868318, 79.10817506193229, 84.73167044595617, 80.37109375, 61.63197642291398, 85.29664660361135, 76.67924528301887, 80.91603053435115, 71.14147909967846, 10.486725663716815, 89.96235884567126, 86.93659281894575, 78.50467289719626, 78.87835384305023, 82.78433809819764, 68.6579722339083, 83.69127516778524, 84.19640461497183, 61.80811808118081, 87.1401151631478, 72.30266230733302, 80.08504606661941, 71.09540636042404, 53.127833182230276, 83.07123034227567, 85.25007964319848, 81.10939907550076, 89.90825688073394, 80.66374740723668, 88.70234347674013, 88.18897637795276, 84.0227088402271, 89.39472322814278, 31.88831741366642, 76.02131438721138, 88.98163606010017, 92.88194444444444, 85.55707139608913, 90.13657056145675, 91.08635097493037, 82.41206030150754, 69.36980947728382]}\n",
      "\n",
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "901667f3cae74e908e7425be17f7053c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Avg time per algorithm (N_CLUSTERS = 6)', max=3), HTML(value=…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "5f7b91aebe8741da8db14ad191ee8018",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Avg acc per algorithm (N_CLUSTERS = 6)', max=3), HTML(value='…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "================================\n",
      "N CLUSTERS =  7\n",
      "================================\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "3786599296f141dfb8561b205c316763",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=3), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average time per image\n",
      "{'K Means': [105.3105354309082, 123.82678985595703, 151.88417434692383, 91.64347648620605, 44.91300582885742, 55.15315532684326, 110.49234867095947, 47.46580123901367, 141.37368202209473, 55.54862022399902, 72.58026599884033, 297.2116947174072, 109.59651470184326, 115.22345542907715, 68.70081424713135, 51.82480812072754, 74.40392971038818, 181.74140453338623, 151.16333961486816, 140.81847667694092, 49.494314193725586, 60.396742820739746, 139.7073745727539, 150.17218589782715, 96.11673355102539, 184.2250108718872, 87.12761402130127, 77.29251384735107, 168.8978672027588, 45.22440433502197, 68.36519241333008, 89.80391025543213, 45.323801040649414, 77.33991146087646, 137.09092140197754, 53.803205490112305, 97.34957218170166, 58.258914947509766, 55.48567771911621, 283.5893392562866, 125.38106441497803, 98.86660575866699, 57.891106605529785, 186.7414951324463], 'Gaussian Mixture': [345.24784088134766, 432.2021484375, 386.5617275238037, 378.4851551055908, 440.11032581329346, 365.62323570251465, 357.56494998931885, 347.4005699157715, 356.1335802078247, 348.6116886138916, 381.58981800079346, 1744.8944807052612, 438.960599899292, 371.3313817977905, 397.5006341934204, 363.79950046539307, 356.6946506500244, 403.6555767059326, 370.41804790496826, 360.47089099884033, 407.54594802856445, 383.60755443573, 374.8542547225952, 421.13592624664307, 329.9802780151367, 453.1973123550415, 394.31097507476807, 397.011399269104, 406.16652965545654, 380.1870822906494, 370.9900140762329, 380.3741693496704, 360.7238292694092, 398.5825777053833, 369.92719173431396, 372.96388149261475, 396.4968681335449, 380.7098865509033, 383.2247018814087, 1742.8832530975342, 367.3357963562012, 351.11982822418213, 440.78617095947266, 375.17497539520264], 'CUDA K Means': [62.99323959350586, 44.068946075439456, 39.174015045166016, 36.18296813964844, 38.284779739379886, 37.715257263183595, 36.534338760375974, 37.66265296936035, 36.92653541564941, 36.81168327331543, 36.35852165222168, 134.36478424072266, 38.99111480712891, 36.77640228271484, 38.78088150024414, 36.76530456542969, 36.364406967163085, 37.129173278808594, 37.04855537414551, 37.56252326965332, 38.79637756347656, 36.795929718017575, 36.87532043457031, 38.36779670715332, 38.57320442199707, 36.13870277404785, 36.409088897705075, 36.15440292358399, 36.17491226196289, 36.22864418029785, 36.31976203918457, 37.44000129699707, 36.40527420043945, 36.40997123718262, 41.32042770385742, 37.47712631225586, 47.397911834716794, 48.279076766967776, 48.61692123413086, 151.44581909179686, 43.31330909729004, 43.641900634765626, 43.56068382263184, 44.001325607299805]}\n",
      "\n",
      "\n",
      "Average accuracy per image\n",
      "{'K Means': [66.84574059861858, 79.95124923826936, 67.88866259334692, 68.39359597049828, 73.45820009136592, 72.99118714359773, 50.626367070988266, 83.79096545615589, 72.07724425887265, 57.17806531115218, 64.41108056671601, 25.574995085512093, 83.27086882453152, 82.1806346623271, 63.719234275296266, 80.58870326173428, 75.05938242280284, 60.34444195929323, 68.28125, 75.18602394047234, 56.22574955908289, 69.91689750692521, 49.44360902255639, 71.03393843725335, 68.73646209386281, 45.71713147410358, 70.62436028659161, 68.88556469708303, 75.5082284607938, 87.85357737104825, 75.88670579229395, 85.02129235326792, 88.5593220338983, 76.25835189309576, 81.61764705882352, 29.529243937232525, 75.31219980787705, 85.5085508550855, 76.97434188603799, 84.72355824286565, 86.15713522180653, 81.69934640522875, 57.56302521008403, 68.01579466929911], 'Gaussian Mixture': [66.22976098689283, 89.29663608562691, 76.42276422764228, 70.85613415710503, 74.83115713642503, 72.59490379615184, 51.365255243371585, 83.36025848142164, 74.66121196624904, 79.75334018499485, 61.60155406863803, 5.534709193245779, 88.12359188928227, 87.83382789317507, 56.45703312530005, 81.01466508125247, 75.47169811320755, 54.384328358208954, 90.45521292217327, 72.38479650184998, 60.75358452817606, 67.19457013574662, 50.951248513674194, 79.49910554561718, 71.96710075394105, 54.26636568848758, 77.34109655507035, 84.64491362763916, 70.18842530282639, 93.25513196480938, 82.1899441340782, 88.3166929409704, 86.68171557562077, 79.05172413793103, 81.39963167587477, 33.003300330033, 75.68523430592397, 87.60991207034373, 80.67226890756302, 84.16643652766271, 88.39378238341969, 80.19966722129783, 80.45801526717557, 69.81499513145083], 'CUDA K Means': [67.88432267884322, 79.60808328230252, 71.72050098879367, 78.86811186496773, 84.56273764258555, 78.35153922542204, 57.2405929304447, 85.17566409597258, 76.36822194199243, 76.17521367521367, 63.893016344725105, 4.503216583273767, 89.80490874764003, 86.26543209876543, 75.57220141489805, 77.44437640334763, 82.72785213511791, 59.93719156572453, 82.84263959390863, 84.14468316562414, 60.36243822075783, 85.01469147894221, 62.450383699391374, 79.22636103151862, 68.84531590413944, 51.72890733056709, 81.97945845004668, 60.263053009167, 76.85924034471753, 88.93780957622455, 80.4352859458208, 85.8348623853211, 88.27586206896552, 77.13401149933658, 82.48175182481752, 33.980582524271846, 76.55954631379961, 88.4083044982699, 79.9349593495935, 82.64316358888267, 89.33129147524247, 90.85794655414908, 58.60271115745568, 75.74349442379183]}\n",
      "\n",
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "1210c36c9d2f4f00a51184cdc7f5ee5e",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Avg time per algorithm (N_CLUSTERS = 7)', max=3), HTML(value=…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "78f941f6633244348b634efe5cbdc635",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Avg acc per algorithm (N_CLUSTERS = 7)', max=3), HTML(value='…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "================================\n",
      "N CLUSTERS =  8\n",
      "================================\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "05197cb741ef4fe481d41f4661e5a4dc",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=3), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average time per image\n",
      "{'K Means': [93.31529140472412, 78.80744934082031, 75.98686218261719, 65.75784683227539, 281.77154064178467, 154.1322946548462, 225.25978088378906, 85.63966751098633, 124.26989078521729, 89.8632287979126, 114.33167457580566, 852.0599603652954, 74.65236186981201, 273.89400005340576, 82.31589794158936, 142.1680212020874, 85.70668697357178, 67.93172359466553, 102.60515213012695, 79.97963428497314, 84.26671028137207, 99.13392066955566, 138.5984182357788, 114.05670642852783, 208.622145652771, 74.14672374725342, 58.556389808654785, 59.90478992462158, 54.36906814575195, 71.10044956207275, 90.9813642501831, 102.05252170562744, 128.5797119140625, 53.03919315338135, 70.2568769454956, 271.22809886932373, 79.07044887542725, 83.5869550704956, 109.64181423187256, 784.7817182540894, 161.2502098083496, 313.6073112487793, 107.61106014251709, 76.85253620147705], 'Gaussian Mixture': [408.19759368896484, 421.8465805053711, 483.64574909210205, 399.59402084350586, 417.3548460006714, 418.2817220687866, 458.8768720626831, 449.11489486694336, 364.9261474609375, 354.5982360839844, 439.62152004241943, 1699.867582321167, 437.1556282043457, 473.1173276901245, 390.66171646118164, 404.84588146209717, 387.92991638183594, 467.86048412323, 406.9399118423462, 430.11295795440674, 529.1629791259766, 476.28142833709717, 460.66551208496094, 425.596284866333, 396.9907522201538, 530.5750608444214, 428.6099910736084, 418.9805507659912, 465.0919437408447, 374.87823963165283, 403.95750999450684, 437.78979778289795, 417.7037715911865, 418.616247177124, 422.30286598205566, 427.3888826370239, 420.39687633514404, 401.349139213562, 408.27951431274414, 1846.8393087387085, 462.72830963134766, 377.18820571899414, 439.03443813323975, 418.29512119293213], 'CUDA K Means': [65.44661445617676, 45.668806838989255, 42.27502784729004, 41.331917953491214, 41.32958335876465, 40.86036911010742, 40.853620529174805, 42.07527465820313, 42.072317886352536, 41.375130844116214, 41.11048889160156, 153.2504653930664, 41.66811103820801, 41.017493057250974, 41.09965438842774, 41.44943962097168, 41.375567245483396, 41.51292839050293, 41.778622436523435, 41.04573097229004, 40.96120376586914, 41.27815246582031, 41.54709053039551, 42.28401145935059, 42.84407997131348, 42.27895164489746, 42.5157886505127, 42.50896873474121, 47.12161026000977, 41.53655319213867, 46.68693466186524, 42.19821968078613, 41.8475341796875, 41.51808853149414, 41.19878959655762, 41.5792236328125, 41.33696517944336, 41.47856101989746, 42.15062065124512, 153.69359436035157, 46.252693557739256, 42.08796691894531, 41.05708618164063, 41.27686882019043]}\n",
      "\n",
      "\n",
      "Average accuracy per image\n",
      "{'K Means': [63.43959071231799, 64.43514644351464, 64.86111111111111, 68.20392722032067, 62.93494704992436, 68.77343331547938, 46.34894661484966, 74.74949899799599, 61.72073342736248, 55.57632398753894, 60.72980017376195, 3.997127812350407, 71.06366755623333, 81.44499178981938, 62.90692342961945, 65.1695486189086, 67.49116607773851, 57.73524720893142, 67.45283018867924, 70.13076393668273, 54.220314735336196, 66.74233825198638, 39.64786436256929, 68.4168012924071, 49.0787269681742, 43.06269270298047, 70.62436028659161, 81.1390284757119, 68.9538043478261, 67.28232189973615, 37.83783783783784, 70.6283280085197, 79.69735182849938, 74.86437613019892, 76.45926876202694, 30.347890451517394, 74.48275862068967, 85.35262206148282, 73.08353386043314, 80.00190730497808, 78.96895137668423, 78.37837837837837, 53.08775731310943, 54.515778019586506], 'Gaussian Mixture': [64.92798754379136, 68.51351351351352, 75.66204287515762, 69.56676769477626, 74.0909090909091, 63.73748609566184, 50.388368850826524, 84.77366255144034, 68.8954265846483, 75.96153846153845, 60.45701849836779, 5.136663524976438, 78.2078853046595, 84.41971383147853, 56.319077366650646, 63.66939146230699, 75.27559055118111, 53.220498354489884, 69.40356312935708, 71.11567419575633, 59.60219478737997, 64.97987349051179, 40.40272815849302, 70.9090909090909, 49.331103678929765, 53.02197802197802, 69.5562435500516, 84.47722899294419, 66.36647038448216, 67.01916721744877, 81.97767145135566, 72.01854193004635, 85.21536670547147, 78.57762359063312, 81.39963167587477, 34.52484742807324, 75.66539923954373, 87.60991207034373, 78.75205254515599, 80.6336859080776, 87.2688853671421, 79.59866220735786, 80.88119590873329, 57.917109458023376], 'CUDA K Means': [74.5702005730659, 78.23383084577115, 69.83277591973244, 71.4461430328589, 84.22264875239924, 71.66055526453641, 48.48729326341267, 85.9375, 74.11402157164869, 72.98630136986301, 59.07692307692308, 2.963322807869808, 89.80490874764003, 83.54838709677419, 66.815144766147, 65.19982038616973, 75.31847133757962, 59.77992364697956, 74.64892830746489, 84.04858299595142, 60.33003300330033, 68.38565022421524, 47.19101123595505, 76.75111773472429, 68.27133479212254, 40.41558441558442, 74.91340920336467, 81.5903775476111, 74.00391900718485, 88.63134657836645, 80.16779305523188, 86.62674650698602, 88.03165182987142, 77.89473684210526, 82.62454434993924, 41.11111111111111, 75.8494031221304, 87.73168578993821, 80.63368897510507, 85.4163820971634, 89.10485933503837, 90.80622347949081, 75.1147842056933, 58.21845174973489]}\n",
      "\n",
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ce736cc7a333474890ecd3a5e32f8a8f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Avg time per algorithm (N_CLUSTERS = 8)', max=3), HTML(value=…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "64340318e12247a2879cac51db78e9fd",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Avg acc per algorithm (N_CLUSTERS = 8)', max=3), HTML(value='…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "================================\n",
      "N CLUSTERS =  9\n",
      "================================\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c481584d62044217888f707283998ad6",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=3), HTML(value='')))"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average time per image\n",
      "{'K Means': [82.67688751220703, 59.458279609680176, 87.55183219909668, 186.8659496307373, 76.54368877410889, 74.67875480651855, 337.151575088501, 327.2664785385132, 84.79278087615967, 115.4510498046875, 131.736421585083, 510.57379245758057, 100.07791519165039, 105.16037940979004, 74.84550476074219, 80.45501708984375, 63.366079330444336, 102.9897689819336, 105.76579570770264, 98.11410903930664, 75.66497325897217, 140.30468463897705, 88.3436918258667, 104.47890758514404, 89.49136734008789, 178.35659980773926, 88.24636936187744, 141.7790651321411, 191.6494369506836, 88.00520896911621, 113.57896327972412, 80.82480430603027, 77.00023651123047, 79.81564998626709, 139.3782377243042, 120.03076076507568, 240.89996814727783, 60.303497314453125, 175.37879943847656, 342.76862144470215, 148.40409755706787, 94.15338039398193, 99.33233261108398, 84.8482608795166], 'Gaussian Mixture': [421.8333959579468, 559.1383695602417, 554.5280694961548, 529.559588432312, 496.43635749816895, 451.00018978118896, 538.6965274810791, 465.463662147522, 534.5117330551147, 441.63568019866943, 455.2156448364258, 1770.5204248428345, 468.6760902404785, 441.4259910583496, 519.3951845169067, 469.1983222961426, 511.6633653640747, 519.0502882003784, 554.6017169952393, 507.0704460144043, 477.1120071411133, 562.865161895752, 468.8304662704468, 414.5848274230957, 392.73133277893066, 558.5987567901611, 532.4848651885986, 565.9448862075806, 502.51076221466064, 414.6671772003174, 511.7412328720093, 483.93518924713135, 494.5108890533447, 513.1458044052124, 492.4354314804077, 400.7486581802368, 424.8981475830078, 435.65354347229004, 428.9798974990845, 2102.598190307617, 432.95443058013916, 441.0555124282837, 471.0918664932251, 445.4364776611328], 'CUDA K Means': [72.29559364318848, 50.53536949157715, 45.77542266845703, 45.854706573486325, 45.95570373535156, 45.60886840820312, 45.72383422851563, 45.92145462036133, 45.77276191711426, 45.6877384185791, 45.76896018981934, 168.40961761474608, 46.17023963928223, 45.69916610717773, 45.649482345581056, 45.75599899291992, 45.654405975341795, 45.873783874511716, 45.71510391235351, 45.71614723205566, 45.7430965423584, 45.7566276550293, 45.79094276428223, 45.909753036499026, 45.791996765136716, 45.425035858154295, 45.71649322509766, 45.679626846313475, 45.74781799316406, 45.61275444030762, 45.76313400268555, 45.874914932250974, 45.7455020904541, 45.84459800720215, 45.63316459655762, 45.809722900390625, 45.62537269592285, 45.847160720825194, 45.78056030273437, 168.37585906982423, 45.51732902526855, 46.00520668029785, 45.55468406677246, 45.68585472106933]}\n",
      "\n",
      "\n",
      "Average accuracy per image\n",
      "{'K Means': [63.654223968565816, 61.93181818181818, 61.29722024233784, 60.51477597712107, 53.653217011995636, 61.81613085166384, 46.19164619164619, 73.48178137651821, 61.681715575620764, 54.030226700251895, 59.41679456259592, 0.0, 68.89061287820016, 76.21527777777779, 62.90692342961945, 60.07923560941505, 66.48793565683646, 51.54541131716596, 67.24409448818898, 61.81545386346586, 42.807558812186656, 57.955232909860854, 39.7523623330075, 65.37190082644628, 43.57638888888889, 41.25654450261781, 69.62351727694688, 59.591509811774124, 62.55789098681866, 67.10700132100396, 35.36585365853659, 65.05221061986225, 79.84790874524715, 58.572173006774364, 73.70491803278688, 36.34615384615385, 73.39642481598318, 74.05900305188199, 69.95420922860161, 78.14845300924806, 77.28894173602853, 71.65775401069519, 52.60869565217391, 51.69162506932889], 'Gaussian Mixture': [59.2292089249493, 65.83850931677019, 72.03667321545514, 69.10001789228842, 64.37810945273633, 63.81322957198443, 48.02751365567469, 84.58367683429515, 63.757643135075035, 52.638270820089005, 59.724349157733535, 3.7463976945244957, 74.31328878990348, 82.9975825946817, 55.76459237819586, 64.07063617840163, 74.05012126111559, 53.289473684210535, 65.76433121019109, 68.01132342533617, 59.05947441217151, 64.35586366262275, 39.79125896934116, 70.75583696082312, 49.331103678929765, 42.959917780061666, 69.08337648886588, 64.47622729029764, 65.10489510489509, 65.32797858099063, 81.49184149184148, 71.77538526493561, 83.4340991535671, 76.33451957295374, 73.28947368421052, 37.3904576436222, 75.38167938931298, 87.60991207034373, 75.81567440295997, 77.57345787118116, 81.62564396107614, 71.42857142857143, 80.91482649842271, 54.752851711026615], 'CUDA K Means': [72.55474452554745, 78.23383084577115, 69.83277591973244, 68.01371098683023, 84.83483483483484, 66.77542142468734, 47.34809997967893, 84.42028985507247, 76.14957049014653, 73.05585980284775, 64.41108056671601, 3.0799078452770705, 84.90566037735849, 83.57487922705315, 64.82820976491863, 62.41979835013749, 75.31847133757962, 53.80472957152892, 67.60895170789163, 84.00541271989175, 60.78750406768631, 68.08988764044945, 48.727925770727325, 71.71439780306002, 68.07888970051133, 48.4472049689441, 65.67481402763018, 60.54097056483692, 68.46232526423456, 69.94182288299935, 80.44588945657223, 86.77536231884058, 87.71228771228772, 77.69467663880334, 82.48175182481752, 46.17834394904459, 76.0890609874153, 87.61061946902655, 85.19900497512438, 84.81325521786891, 89.18640576725025, 86.90292758089369, 74.14272474513439, 58.21845174973489]}\n",
      "\n",
      "\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "28b8bc4175844893b440017364dd094b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Avg time per algorithm (N_CLUSTERS = 9)', max=3), HTML(value=…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2afacbe5973e4474b80f2c7fd942cdf2",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, description='Avg acc per algorithm (N_CLUSTERS = 9)', max=3), HTML(value='…"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Total average time (msecs)\n",
      "[[31.912983547557484, 250.83396326411855, 21.041653849861838], [49.02191487225619, 276.29934332587504, 25.15742429386485], [62.036899003115565, 318.4749792922627, 31.15182054693049], [83.6683452129364, 370.0789142738689, 36.35587107484991], [106.47322210398588, 445.14880017800766, 44.15090852217241], [148.76691157167608, 489.1648254611275, 47.67810733101585], [132.9218176278201, 549.5258309624412, 52.0404901938005]]\n",
      "\n",
      "\n",
      "95% Confidence interval of time (msecs)\n",
      "[[[24.532093640576154, 39.29387345453881], [203.78577688001957, 297.88214964821753], [17.878547082805134, 24.20476061691854]], [[38.993279457227885, 59.050550287284494], [221.6862481774976, 330.9124384742525], [21.327039431014835, 28.987809156714864]], [[48.66103445103141, 75.41276355519972], [261.235022397772, 375.71493618675333], [26.485903364791945, 35.81773772906904]], [[70.83033192800929, 96.50635849786352], [309.16002364362447, 430.99780490411337], [30.994995101799425, 41.71674704790039]], [[88.8850792710094, 124.06136493696235], [358.5757404464477, 531.7218599095677], [37.40734983684344, 50.89446720750138]], [[100.16934807519448, 197.36447506815767], [403.1149282135146, 575.2147227087405], [40.56598206371088, 54.79023259832083]], [[105.62622164905824, 160.21741360658197], [455.74491986120483, 643.3067420636776], [44.22588543455567, 59.85509495304533]]]\n",
      "\n",
      "\n",
      "Total average accuracy (%)\n",
      "[[66.71818937455167, 61.813777480807, 72.9656405790198], [79.70465936373454, 80.37642535127215, 81.68719896744157], [78.16947927119344, 78.76085936717386, 80.54547968258625], [75.169099582798, 77.12973543265156, 77.78574118614256], [70.1952256138125, 73.18949324176246, 74.29504406897168], [63.546379976145495, 68.21476009451432, 72.102534771699], [59.5488347230377, 65.09543809696902, 70.4339260276681]]\n",
      "\n",
      "\n",
      "95% Confidence interval of accuracy (%)\n",
      "[[[57.258805212600734, 76.1775735365026], [53.17259982631046, 70.45495513530354], [64.96660331190681, 80.9646778461328]], [[74.75980273293264, 84.64951599453643], [76.18514777012486, 84.56770293241944], [77.10254813483247, 86.27184980005067]], [[73.7131498223769, 82.62580872000997], [74.19298659294121, 83.32873214140652], [76.39513729722162, 84.69582206795087]], [[70.70933584367361, 79.62886332192238], [72.42572399437408, 81.83374687092903], [73.19251189955342, 82.3789704727317]], [[65.95420257028695, 74.43624865733804], [68.23435097262592, 78.14463551089901], [69.47301785852176, 79.1170702794216]], [[58.89921127860413, 68.19354867368686], [63.47088455635071, 72.95863563267793], [67.10197750186327, 77.10309204153472]], [[55.14451785702922, 63.95315158904618], [60.4953654942953, 69.69551069964274], [65.67553013692284, 75.19232191841334]]]\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "N_CLUSTERS_RANGE = (3, 10)\n",
    "t_avg_accuracy = []\n",
    "conf95_accuracy = []\n",
    "\n",
    "t_avg_time = []\n",
    "conf95_time = []\n",
    "\n",
    "CONF_INT_COEFF = 1.0 / sqrt(N) * t(0.05, N - 1)\n",
    "\n",
    "avg_time_per_n_clusters = []\n",
    "avg_acc_per_n_clusters = []\n",
    "\n",
    "for N_CLUSTERS in tqdm(range(N_CLUSTERS_RANGE[0], N_CLUSTERS_RANGE[1]), desc='Collecting data'):\n",
    "    print('================================')\n",
    "    print('N CLUSTERS = ', N_CLUSTERS)\n",
    "    print('================================')\n",
    "    time_t, accuracy_t = {}, {}\n",
    "    for alg in tqdm(ALGS.keys()):\n",
    "        avg_time, avg_accuracy = profile(ALGS[alg], PREP, N_CLUSTERS)\n",
    "        time_t[alg] = avg_time\n",
    "        accuracy_t[alg] = avg_accuracy\n",
    "\n",
    "    print('Average time per image')\n",
    "    print(time_t)\n",
    "    print('\\n')\n",
    "\n",
    "    print('Average accuracy per image')\n",
    "    print(accuracy_t)\n",
    "    print('\\n')\n",
    "\n",
    "    \n",
    "    t_avg_time_alg = []\n",
    "    t_avg_accuracy_alg = []\n",
    "    conf95_time_alg = []\n",
    "    conf95_accuracy_alg = []\n",
    "    \n",
    "    avg_time_per_alg = []\n",
    "    for alg, time_per_image in tqdm(time_t.items(), desc='Avg time per algorithm (N_CLUSTERS = %d)' % N_CLUSTERS):\n",
    "        avg_time_per_alg.append(time_per_image)\n",
    "        t_avg_time_ = np.mean(time_per_image)\n",
    "        t_std_time_ = np.std(time_per_image)\n",
    "        t_avg_time_alg.append(t_avg_time_)\n",
    "#         t_std_time[i].append(t_std_time_)\n",
    "        conf95_time_alg.append(\n",
    "            [\n",
    "                t_avg_time_ - t_std_time_ * CONF_INT_COEFF, \n",
    "                t_avg_time_ + t_std_time_ * CONF_INT_COEFF\n",
    "            ]\n",
    "        )\n",
    "    avg_time_per_n_clusters.append(avg_time_per_alg)\n",
    "\n",
    "    avg_acc_per_alg = []\n",
    "    for alg, accuracy_per_image in tqdm(accuracy_t.items(), desc='Avg acc per algorithm (N_CLUSTERS = %d)' % N_CLUSTERS):\n",
    "        avg_acc_per_alg.append(accuracy_per_image)\n",
    "        t_avg_accuracy_ = np.mean(accuracy_per_image)\n",
    "        t_std_accuracy_ = np.std(accuracy_per_image)\n",
    "        t_avg_accuracy_alg.append(t_avg_accuracy_)\n",
    "#         t_std_accuracy[i].append(t_std_accuracy_)\n",
    "        conf95_accuracy_alg.append(\n",
    "            [\n",
    "                t_avg_accuracy_ - t_std_accuracy_ * CONF_INT_COEFF, \n",
    "                t_avg_accuracy_ + t_std_accuracy_ * CONF_INT_COEFF\n",
    "            ]\n",
    "        )\n",
    "    avg_acc_per_n_clusters.append(avg_time_per_alg)\n",
    "    \n",
    "    t_avg_time.append(t_avg_time_alg)\n",
    "    t_avg_accuracy.append(t_avg_accuracy_alg)\n",
    "    conf95_time.append(conf95_time_alg)\n",
    "    conf95_accuracy.append(conf95_accuracy_alg)\n",
    "\n",
    "print('Total average time (msecs)')\n",
    "print(t_avg_time)\n",
    "print('\\n')\n",
    "\n",
    "# print('Total std of time (msecs)^2')\n",
    "# print(t_std_time)\n",
    "# print('\\n')\n",
    "\n",
    "print('95% Confidence interval of time (msecs)')\n",
    "print(conf95_time)\n",
    "print('\\n')\n",
    "\n",
    "print('Total average accuracy (%)')\n",
    "print(t_avg_accuracy)\n",
    "print('\\n')\n",
    "\n",
    "#     print('Total std of accuracy (%)')\n",
    "#     print(t_std_accuracy)\n",
    "#     print('\\n')\n",
    "\n",
    "print('95% Confidence interval of accuracy (%)')\n",
    "print(conf95_accuracy)\n",
    "print('\\n')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "===========================\n",
      "-----> K Means <-----\n",
      "\n",
      "\n",
      "[Average time]\n",
      "[ 31.91298355  49.02191487  62.036899    83.66834521 106.4732221\n",
      " 148.76691157 132.92181763]\n",
      "\n",
      "\n",
      "[Average accuracy]\n",
      "[66.71818937 79.70465936 78.16947927 75.16909958 70.19522561 63.54637998\n",
      " 59.54883472]\n",
      "\n",
      "\n",
      "[Conf 95% time]\n",
      "[[ 24.53209364  39.29387345]\n",
      " [ 38.99327946  59.05055029]\n",
      " [ 48.66103445  75.41276356]\n",
      " [ 70.83033193  96.5063585 ]\n",
      " [ 88.88507927 124.06136494]\n",
      " [100.16934808 197.36447507]\n",
      " [105.62622165 160.21741361]]\n",
      "\n",
      "\n",
      "[Conf 95% accuracy]\n",
      "[[57.25880521 76.17757354]\n",
      " [74.75980273 84.64951599]\n",
      " [73.71314982 82.62580872]\n",
      " [70.70933584 79.62886332]\n",
      " [65.95420257 74.43624866]\n",
      " [58.89921128 68.19354867]\n",
      " [55.14451786 63.95315159]]\n",
      "\n",
      "\n",
      "===========================\n",
      "-----> Gaussian Mixture <-----\n",
      "\n",
      "\n",
      "[Average time]\n",
      "[250.83396326 276.29934333 318.47497929 370.07891427 445.14880018\n",
      " 489.16482546 549.52583096]\n",
      "\n",
      "\n",
      "[Average accuracy]\n",
      "[61.81377748 80.37642535 78.76085937 77.12973543 73.18949324 68.21476009\n",
      " 65.0954381 ]\n",
      "\n",
      "\n",
      "[Conf 95% time]\n",
      "[[203.78577688 297.88214965]\n",
      " [221.68624818 330.91243847]\n",
      " [261.2350224  375.71493619]\n",
      " [309.16002364 430.9978049 ]\n",
      " [358.57574045 531.72185991]\n",
      " [403.11492821 575.21472271]\n",
      " [455.74491986 643.30674206]]\n",
      "\n",
      "\n",
      "[Conf 95% accuracy]\n",
      "[[53.17259983 70.45495514]\n",
      " [76.18514777 84.56770293]\n",
      " [74.19298659 83.32873214]\n",
      " [72.42572399 81.83374687]\n",
      " [68.23435097 78.14463551]\n",
      " [63.47088456 72.95863563]\n",
      " [60.49536549 69.6955107 ]]\n",
      "\n",
      "\n",
      "===========================\n",
      "-----> CUDA K Means <-----\n",
      "\n",
      "\n",
      "[Average time]\n",
      "[21.04165385 25.15742429 31.15182055 36.35587107 44.15090852 47.67810733\n",
      " 52.04049019]\n",
      "\n",
      "\n",
      "[Average accuracy]\n",
      "[72.96564058 81.68719897 80.54547968 77.78574119 74.29504407 72.10253477\n",
      " 70.43392603]\n",
      "\n",
      "\n",
      "[Conf 95% time]\n",
      "[[17.87854708 24.20476062]\n",
      " [21.32703943 28.98780916]\n",
      " [26.48590336 35.81773773]\n",
      " [30.9949951  41.71674705]\n",
      " [37.40734984 50.89446721]\n",
      " [40.56598206 54.7902326 ]\n",
      " [44.22588543 59.85509495]]\n",
      "\n",
      "\n",
      "[Conf 95% accuracy]\n",
      "[[64.96660331 80.96467785]\n",
      " [77.10254813 86.2718498 ]\n",
      " [76.3951373  84.69582207]\n",
      " [73.1925119  82.37897047]\n",
      " [69.47301786 79.11707028]\n",
      " [67.1019775  77.10309204]\n",
      " [65.67553014 75.19232192]]\n"
     ]
    }
   ],
   "source": [
    "from pylab import rcParams\n",
    "rcParams['figure.figsize'] = 10, 10\n",
    "rcParams['legend.fontsize'] = 15\n",
    "rcParams['xtick.labelsize'] = 15\n",
    "rcParams['ytick.labelsize'] = 15\n",
    "\n",
    "res_time = np.asarray(t_avg_time)\n",
    "res_acc = np.asarray(t_avg_accuracy)\n",
    "conf95_time_res = np.asarray(conf95_time)\n",
    "conf95_acc_res = np.asarray(conf95_accuracy)\n",
    "\n",
    "avg_time_per_n_clusters = np.asarray(avg_time_per_n_clusters)\n",
    "avg_acc_per_n_clusters = np.asarray(avg_acc_per_n_clusters)\n",
    "\n",
    "X = np.arange(N_CLUSTERS_RANGE[0], N_CLUSTERS_RANGE[1])\n",
    "plt.title('Average time (msecs)', fontsize=18, fontweight='bold')\n",
    "plt.xticks(X)\n",
    "plt.xlabel('n clusters', fontweight=\"bold\", fontsize=16)\n",
    "plt.ylabel('msecs', fontweight=\"bold\", fontsize=16)\n",
    "plt.ylim(0, np.max(res_time) * 1.1)\n",
    "# plt.yscale('log', basey=10)\n",
    "for i, alg in zip(range(len(ALGS.keys())), ALGS.keys()):\n",
    "    # ax.set_xticklabels(['K=%d' % k for k in range(3, 6)])\n",
    "    plt.plot(X, res_time[:,i], label=alg)\n",
    "#     plt.fill_between(X, conf95_time_res[:,i,0], conf95_time_res[:,i,1], alpha=0.15)\n",
    "    # plt.axis('equal')\n",
    "plt.legend()\n",
    "plt.grid()\n",
    "plt.show()\n",
    "\n",
    "\n",
    "plt.title('Average accuracy (%)', fontsize=18, fontweight='bold')\n",
    "plt.xticks(X)\n",
    "plt.xlabel('n clusters', fontweight=\"bold\", fontsize=16)\n",
    "plt.ylabel('%', fontweight=\"bold\", fontsize=16)\n",
    "plt.ylim(0, np.max(res_acc) * 1.1)\n",
    "# plt.yscale('log', basey=10)\n",
    "for i, alg in zip(range(len(ALGS.keys())), ALGS.keys()):\n",
    "    # ax.set_xticklabels(['K=%d' % k for k in range(3, 6)])\n",
    "    plt.plot(X, res_acc[:,i], label=alg)\n",
    "#     plt.fill_between(X, conf95_acc_res[:,i,0], conf95_acc_res[:,i,1], alpha=0.15)\n",
    "    # plt.axis('equal')\n",
    "plt.legend()\n",
    "plt.grid()\n",
    "plt.show()\n",
    "\n",
    "import matplotlib\n",
    "palegreen = matplotlib.colors.colorConverter.to_rgb('#8CFF6F')\n",
    "paleblue = matplotlib.colors.colorConverter.to_rgb('#708DFF')\n",
    "    \n",
    "# for i, alg in zip(range(len(ALGS.keys())), ALGS.keys()):\n",
    "#     fig = plt.figure(1, figsize=(9, 9))\n",
    "#     ax = fig.add_subplot(111)\n",
    "    \n",
    "#     ax.set_title('%s - Time (msecs)' % alg)\n",
    "#     bp = ax.boxplot(avg_time_per_n_clusters[:,i,:].T)\n",
    "#     ax.set_xticklabels(list(range(N_CLUSTERS_RANGE[0], N_CLUSTERS_RANGE[1])))\n",
    "#     plt.show()\n",
    "\n",
    "#     fig = plt.figure(1, figsize=(9, 9))\n",
    "#     ax = fig.add_subplot(111)\n",
    "    \n",
    "#     ax.set_title('%s - Accuracy (%%)' % alg)\n",
    "#     bp = ax.boxplot(avg_acc_per_n_clusters[:,i,:].T)\n",
    "#     ax.set_xticklabels(list(range(N_CLUSTERS_RANGE[0], N_CLUSTERS_RANGE[1])))\n",
    "#     plt.show()\n",
    "\n",
    "for i, alg in zip(range(len(ALGS.keys())), ALGS.keys()):\n",
    "    print(\"\\n\")\n",
    "    print(\"===========================\")\n",
    "    print('----->', alg, '<-----')\n",
    "    print(\"\\n\")\n",
    "    print('[Average time]')\n",
    "    print(res_time[:,i])\n",
    "    print(\"\\n\")\n",
    "    print('[Average accuracy]')\n",
    "    print(res_acc[:,i])\n",
    "    print(\"\\n\")\n",
    "    print('[Conf 95% time]')\n",
    "    print(conf95_time_res[:,i,:])\n",
    "    print(\"\\n\")\n",
    "    print('[Conf 95% accuracy]')\n",
    "    print(conf95_acc_res[:,i,:])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "\n",
      "===========================\n",
      "-----> K Means <-----\n",
      "\n",
      "\n",
      "[Average time]\n",
      "[ 31.91298355  49.02191487  62.036899    83.66834521 106.4732221\n",
      " 148.76691157 132.92181763]\n",
      "\n",
      "\n",
      "[Average accuracy]\n",
      "[66.71818937 79.70465936 78.16947927 75.16909958 70.19522561 63.54637998\n",
      " 59.54883472]\n",
      "\n",
      "\n",
      "[Conf 95% time]\n",
      "[[ 24.53209364  39.29387345]\n",
      " [ 38.99327946  59.05055029]\n",
      " [ 48.66103445  75.41276356]\n",
      " [ 70.83033193  96.5063585 ]\n",
      " [ 88.88507927 124.06136494]\n",
      " [100.16934808 197.36447507]\n",
      " [105.62622165 160.21741361]]\n",
      "\n",
      "\n",
      "[Conf 95% accuracy]\n",
      "[[57.25880521 76.17757354]\n",
      " [74.75980273 84.64951599]\n",
      " [73.71314982 82.62580872]\n",
      " [70.70933584 79.62886332]\n",
      " [65.95420257 74.43624866]\n",
      " [58.89921128 68.19354867]\n",
      " [55.14451786 63.95315159]]\n"
     ]
    }
   ],
   "source": [
    "from pylab import rcParams\n",
    "# rcParams['title.fontsize'] = 18\n",
    "# rcParams['title.fontname'] = \"Times New Roman Bold\"\n",
    "rcParams['figure.figsize'] = 10, 10\n",
    "rcParams['legend.fontsize'] = 17\n",
    "rcParams['xtick.labelsize'] = 15\n",
    "rcParams['ytick.labelsize'] = 15\n",
    "\n",
    "res_time = np.asarray(t_avg_time)\n",
    "res_acc = np.asarray(t_avg_accuracy)\n",
    "conf95_time_res = np.asarray(conf95_time)\n",
    "conf95_acc_res = np.asarray(conf95_accuracy)\n",
    "\n",
    "avg_time_per_n_clusters = np.asarray(avg_time_per_n_clusters)\n",
    "avg_acc_per_n_clusters = np.asarray(avg_acc_per_n_clusters)\n",
    "\n",
    "X = np.arange(N_CLUSTERS_RANGE[0], N_CLUSTERS_RANGE[1])\n",
    "plt.title('Average time (msecs)', fontsize=18, fontweight=\"bold\")\n",
    "plt.xticks(X)\n",
    "plt.xlabel('n clusters', fontsize=18, fontweight=\"bold\")\n",
    "plt.ylabel('msecs', fontsize=18, fontweight=\"bold\")\n",
    "plt.ylim(0, np.max(conf95_time_res[:, [0, 2], 1]) * 1.1)\n",
    "# plt.yscale('log', basey=10)\n",
    "for i, alg in zip(range(len(ALGS.keys())), ALGS.keys()):\n",
    "    # ax.set_xticklabels(['K=%d' % k for k in range(3, 6)])\n",
    "    if i == 0 or i == 4:\n",
    "        plt.plot(X, res_time[:,i], label=alg)\n",
    "        plt.fill_between(X, conf95_time_res[:,i,0], conf95_time_res[:,i,1], alpha=0.08)\n",
    "    # plt.axis('equal')\n",
    "plt.legend()\n",
    "plt.grid()\n",
    "plt.show()\n",
    "\n",
    "\n",
    "plt.title('Average accuracy (%)', fontsize=18, fontname=\"Times New Roman Bold\", fontweight=\"bold\")\n",
    "plt.xticks(X)\n",
    "plt.xlabel('n clusters', fontsize=18, fontweight=\"bold\")\n",
    "plt.ylabel('%', fontsize=18, fontweight=\"bold\")\n",
    "plt.ylim(0, np.max(conf95_acc_res[:, [0, 2], 1]) * 1.1)\n",
    "# plt.yscale('log', basey=10)\n",
    "for i, alg in zip(range(len(ALGS.keys())), ALGS.keys()):\n",
    "    if i == 0 or i == 4:\n",
    "    # ax.set_xticklabels(['K=%d' % k for k in range(3, 6)])\n",
    "        plt.plot(X, res_acc[:,i], label=alg)\n",
    "        plt.fill_between(X, conf95_acc_res[:,i,0], conf95_acc_res[:,i,1], alpha=0.08)\n",
    "    # plt.axis('equal')\n",
    "plt.legend()\n",
    "plt.grid()\n",
    "plt.show()\n",
    "\n",
    "import matplotlib\n",
    "palegreen = matplotlib.colors.colorConverter.to_rgb('#8CFF6F')\n",
    "paleblue = matplotlib.colors.colorConverter.to_rgb('#708DFF')\n",
    "    \n",
    "# for i, alg in zip(range(len(ALGS.keys())), ALGS.keys()):\n",
    "#     if i == 0 or i == 4:\n",
    "#         fig = plt.figure(1, figsize=(9, 9))\n",
    "#         ax = fig.add_subplot(111)\n",
    "\n",
    "#         ax.set_title('%s - Time (msecs)' % alg)\n",
    "#         bp = ax.boxplot(avg_time_per_n_clusters[:,i,:].T)\n",
    "#         ax.set_xticklabels(list(range(N_CLUSTERS_RANGE[0], N_CLUSTERS_RANGE[1])))\n",
    "#         plt.show()\n",
    "\n",
    "#         fig = plt.figure(1, figsize=(9, 9))\n",
    "#         ax = fig.add_subplot(111)\n",
    "\n",
    "#         ax.set_title('%s - Accuracy (%%)' % alg)\n",
    "#         bp = ax.boxplot(avg_acc_per_n_clusters[:,i,:].T)\n",
    "#         ax.set_xticklabels(list(range(N_CLUSTERS_RANGE[0], N_CLUSTERS_RANGE[1])))\n",
    "#         plt.show()\n",
    "\n",
    "for i, alg in zip(range(len(ALGS.keys())), ALGS.keys()):\n",
    "    if i == 0 or i == 4:\n",
    "        print(\"\\n\")\n",
    "        print(\"===========================\")\n",
    "        print('----->', alg, '<-----')\n",
    "        print(\"\\n\")\n",
    "        print('[Average time]')\n",
    "        print(res_time[:,i])\n",
    "        print(\"\\n\")\n",
    "        print('[Average accuracy]')\n",
    "        print(res_acc[:,i])\n",
    "        print(\"\\n\")\n",
    "        print('[Conf 95% time]')\n",
    "        print(conf95_time_res[:,i,:])\n",
    "        print(\"\\n\")\n",
    "        print('[Conf 95% accuracy]')\n",
    "        print(conf95_acc_res[:,i,:])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Time\n",
      "\n",
      "                       3       4       5       6       7       8       9\n",
      "K Means            31.91   49.02   62.04   83.67  106.47  148.77  132.92\n",
      "Gaussian Mixture  250.83  276.30  318.47  370.08  445.15  489.16  549.53\n",
      "CUDA K Means       21.04   25.16   31.15   36.36   44.15   47.68   52.04\n",
      "\n",
      "\n",
      "Accuracy\n",
      "\n",
      "                      3      4      5      6      7      8      9\n",
      "K Means           66.72  79.70  78.17  75.17  70.20  63.55  59.55\n",
      "Gaussian Mixture  61.81  80.38  78.76  77.13  73.19  68.21  65.10\n",
      "CUDA K Means      72.97  81.69  80.55  77.79  74.30  72.10  70.43\n",
      "\n",
      "\n",
      "Time (95% confidence interval)\n",
      "\n",
      "                               3               4               5  \\\n",
      "K Means           [24.53, 39.29]  [38.99, 59.05]  [48.66, 75.41]   \n",
      "Gaussian Mixture  [203.79, 297.8  [221.69, 330.9  [261.24, 375.7   \n",
      "CUDA K Means      [17.88, 24.20]  [21.33, 28.99]  [26.49, 35.82]   \n",
      "\n",
      "                               6               7               8  \\\n",
      "K Means           [70.83, 96.51]  [88.89, 124.06  [100.17, 197.3   \n",
      "Gaussian Mixture  [309.16, 431.0  [358.58, 531.7  [403.11, 575.2   \n",
      "CUDA K Means      [30.99, 41.72]  [37.41, 50.89]  [40.57, 54.79]   \n",
      "\n",
      "                               9  \n",
      "K Means           [105.63, 160.2  \n",
      "Gaussian Mixture  [455.74, 643.3  \n",
      "CUDA K Means      [44.23, 59.86]  \n",
      "\n",
      "\n",
      "Accuracy (95% confidence interval)\n",
      "\n",
      "                               3               4               5  \\\n",
      "K Means           [57.26, 76.18]  [74.76, 84.65]  [73.71, 82.63]   \n",
      "Gaussian Mixture  [53.17, 70.45]  [76.19, 84.57]  [74.19, 83.33]   \n",
      "CUDA K Means      [64.97, 80.96]  [77.10, 86.27]  [76.40, 84.70]   \n",
      "\n",
      "                               6               7               8  \\\n",
      "K Means           [70.71, 79.63]  [65.95, 74.44]  [58.90, 68.19]   \n",
      "Gaussian Mixture  [72.43, 81.83]  [68.23, 78.14]  [63.47, 72.96]   \n",
      "CUDA K Means      [73.19, 82.38]  [69.47, 79.12]  [67.10, 77.10]   \n",
      "\n",
      "                               9  \n",
      "K Means           [55.14, 63.95]  \n",
      "Gaussian Mixture  [60.50, 69.70]  \n",
      "CUDA K Means      [65.68, 75.19]  \n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "df_time = pd.DataFrame(np.round(res_time.T, decimals=2), columns=[str(k) for k in range(N_CLUSTERS_RANGE[0], N_CLUSTERS_RANGE[1])])\n",
    "df_time.index = ALGS.keys()\n",
    "df_acc = pd.DataFrame(np.round(res_acc.T, decimals=2), columns=[str(k) for k in range(N_CLUSTERS_RANGE[0], N_CLUSTERS_RANGE[1])])\n",
    "df_acc.index = ALGS.keys()\n",
    "\n",
    "conf95_time_res[conf95_time_res < 0] = 0\n",
    "conf95_acc_res[conf95_acc_res < 0] = 0\n",
    "\n",
    "# Conf 95 time\n",
    "df_conf_time = np.apply_along_axis(\n",
    "    lambda x: '[%.2f, %.2f]' % (x[0], x[1]), \n",
    "    2, \n",
    "    np.round(conf95_time_res, decimals=2)\n",
    ").T\n",
    "df_conf_time = pd.DataFrame(df_conf_time, columns=[str(k) for k in range(N_CLUSTERS_RANGE[0], N_CLUSTERS_RANGE[1])])\n",
    "df_conf_time.index = ALGS.keys()\n",
    "\n",
    "# Conf 95 accuracy\n",
    "df_conf_acc = np.apply_along_axis(\n",
    "    lambda x: '[%.2f, %.2f]' % (x[0], x[1]), \n",
    "    2, \n",
    "    np.round(conf95_acc_res, decimals=2)\n",
    ").T\n",
    "df_conf_acc = pd.DataFrame(df_conf_acc, columns=[str(k) for k in range(N_CLUSTERS_RANGE[0], N_CLUSTERS_RANGE[1])])\n",
    "df_conf_acc.index = ALGS.keys()\n",
    "\n",
    "print('Time\\n')\n",
    "print(df_time)\n",
    "print('\\n')\n",
    "\n",
    "print('Accuracy\\n')\n",
    "print(df_acc)\n",
    "print('\\n')\n",
    "\n",
    "print('Time (95% confidence interval)\\n')\n",
    "print(df_conf_time)\n",
    "print('\\n')\n",
    "\n",
    "print('Accuracy (95% confidence interval)\\n')\n",
    "print(df_conf_acc)\n",
    "\n",
    "writer = pd.ExcelWriter('benchmark_max.xlsx')\n",
    "df_time.to_excel(writer, 'Time')\n",
    "df_acc.to_excel(writer, 'Accuracy')\n",
    "df_conf_time.to_excel(writer, 'Time - 95')\n",
    "df_conf_acc.to_excel(writer, 'Accuracy - 95')\n",
    "writer.save()"
   ]
  }
 ],
 "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
}