SketchBasedShapeRetrieval/IPython/08 Compute Matches from Input Sketch (Cython).ipynb

{
 "metadata": {
  "name": "08 Compute Matches from Input Sketch (Cython)"
 },
 "nbformat": 3,
 "nbformat_minor": 0,
 "worksheets": [
  {
   "cells": [
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "%load_ext cythonmagic\n",
      "\n",
      "import numpy\n",
      "import math\n",
      "import scipy\n",
      "import scipy.io\n",
      "import scipy.misc\n",
      "import scipy.cluster\n",
      "import scipy.cluster.vq\n",
      "import matplotlib\n",
      "import matplotlib.pyplot\n",
      "import os\n",
      "import IPython\n",
      "import IPython.parallel\n",
      "import itertools\n",
      "import random\n",
      "import sklearn\n",
      "import sklearn.decomposition\n",
      "import sklearn.manifold\n",
      "import sklearn.cluster\n",
      "import sklearn.feature_extraction\n",
      "import sklearn.feature_extraction.text\n",
      "\n",
      "base_path = \"/u/mlrobert/code/local/2013_drawing_assistant/data/\""
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 312
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "img_path    = base_path+\"renderings/bunny2/apparent_ridges.num_latitude_lines=20.num_longitude_lines=20/\"\n",
      "filter_path = base_path+\"filters/gabor/num_thetas=08/\"\n",
      "\n",
      "local_feature_cluster_centroids_path = \\\n",
      "    base_path+\"local_feature_cluster_centroids/bunny2/\"+ \\\n",
      "    \"apparent_ridges.num_latitude_lines=20.num_longitude_lines=20.gabor.num_thetas=08.galif.patch_width=15.num_samples=32.num_tiles=04.num_samples=1000000.k=5000/\"\n",
      "\n",
      "global_features_path = \\\n",
      "    base_path+\"global_features/bunny2/\"+ \\\n",
      "    \"apparent_ridges.num_latitude_lines=20.num_longitude_lines=20.gabor.num_thetas=08.galif.patch_width=15.num_samples=32.num_tiles=04.num_samples=1000000.k=5000/\"\n",
      "\n",
      "img_names_exts                                = sorted(os.listdir(img_path))\n",
      "local_feature_cluster_centroids_path_name_ext = local_feature_cluster_centroids_path+\"local_feature_cluster_centroids.mat\"\n",
      "global_features_names_exts                    = sorted(os.listdir(global_features_path))\n",
      "num_thetas                                    = len(sorted(os.listdir(filter_path)))\n",
      "num_images                                    = len(img_names_exts)\n",
      "codebook_size                                 = 5000"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 313
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "filter_path = base_path+\"filters/gabor/num_thetas=08/\"\n",
      "\n",
      "tmp = []\n",
      "\n",
      "for theta_index in range(num_thetas):\n",
      "    \n",
      "    current_g_mat = scipy.io.loadmat(filter_path+\"theta=%02d.mat\" % theta_index)\n",
      "    current_g     = current_g_mat[\"g\"]\n",
      "    \n",
      "    tmp.append(current_g)\n",
      "\n",
      "g = numpy.array(tmp)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 314
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "local_feature_cluster_centroids_mat = scipy.io.loadmat(local_feature_cluster_centroids_path_name_ext)\n",
      "local_feature_cluster_centroids     = local_feature_cluster_centroids_mat[\"local_feature_cluster_centroids\"]"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 315
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "all_global_features = []\n",
      "\n",
      "for global_features_name_ext in global_features_names_exts:\n",
      "    global_features_mat = scipy.io.loadmat(global_features_path+global_features_name_ext)\n",
      "    global_features     = global_features_mat[\"global_features\"].ravel()\n",
      "    all_global_features.append(global_features)\n",
      "\n",
      "all_global_features_2d        = numpy.array(all_global_features)\n",
      "tf_idf_transformer            = sklearn.feature_extraction.text.TfidfTransformer()\n",
      "tf_idf_all_global_features_2d = numpy.array(tf_idf_transformer.fit_transform(all_global_features_2d).todense()).squeeze()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 316
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "img_index = 124\n",
      "img       = scipy.misc.imread(img_path+img_names_exts[img_index]).astype(numpy.float32)[:,:,0]\n",
      "img       = (255.0 - img) / 255.0\n",
      "\n",
      "figsize(4,4)\n",
      "imshow(img, cmap=\"gray\");\n",
      "title(img_names_exts[img_index])\n",
      "colorbar();"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD1CAYAAACfkyOsAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nbH4in2LzE/kUm5/Ip9j8RD7F5ifyqd+Q8eQ/NLLcSAAAAABJRU5ErkJggg==\n"
      }
     ],
     "prompt_number": 331
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "num_samples_width                = 32\n",
      "patch_width_as_fraction_of_width = 0.15\n",
      "use_squared_filter_response      = True\n",
      "\n",
      "num_samples_height               = int((float(img.shape[0])*float(num_samples_width))/float(img.shape[1]))\n",
      "samples_y                        = numpy.linspace(0, img.shape[0]-1, num_samples_height+2).astype(numpy.int32)\n",
      "samples_x                        = numpy.linspace(0, img.shape[1]-1, num_samples_width+2).astype(numpy.int32)\n",
      "patch_size                       = int(img.shape[1] * patch_width_as_fraction_of_width)\n",
      "\n",
      "if patch_size % 2 == 0:\n",
      "    patch_size = patch_size + 1\n",
      "\n",
      "patch_half_size = int((patch_size - 1) / 2)\n",
      "\n",
      "filter_responses_padded = zeros((num_thetas, img.shape[0]+patch_size-1, img.shape[1]+patch_size-1), dtype=float32)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 320
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def compute_filter_responses_padded():\n",
      "    for theta_index in range(num_thetas):\n",
      "        \n",
      "        F_img                              = numpy.fft.fft2(img)\n",
      "        F_img_centered                     = numpy.fft.fftshift(F_img)\n",
      "        F_img_centered_filtered            = F_img_centered * g[theta_index]\n",
      "        F_img_centered_filtered_uncentered = numpy.fft.ifftshift(F_img_centered_filtered)\n",
      "        filter_response                    = numpy.fft.ifft2(F_img_centered_filtered_uncentered)\n",
      "    \n",
      "        if use_squared_filter_response:\n",
      "            filter_response = abs(filter_response)*abs(filter_response)\n",
      "        else:\n",
      "            filter_response = abs(filter_response)\n",
      "        \n",
      "        filter_responses_padded[theta_index,patch_half_size:img.shape[0]+patch_half_size,patch_half_size:img.shape[1]+patch_half_size] = \\\n",
      "            filter_response"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 321
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "compute_filter_responses_padded()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 322
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "%%cython\n",
      "\n",
      "import  numpy\n",
      "cimport numpy\n",
      "\n",
      "import scipy\n",
      "import scipy.cluster\n",
      "import scipy.cluster.vq\n",
      "\n",
      "ctypedef numpy.float32_t FLOAT32_DTYPE_t\n",
      "ctypedef numpy.int32_t   INT32_DTYPE_t\n",
      "\n",
      "cdef inline float _mean(numpy.ndarray[FLOAT32_DTYPE_t, ndim=2] a):\n",
      "    cdef int   y = 0\n",
      "    cdef int   x = 0\n",
      "    cdef float s = 0\n",
      "    cdef float n = a.shape[0]*a.shape[1]\n",
      "    for y in range(a.shape[0]):\n",
      "        for x in range(a.shape[1]):\n",
      "            s = s+a[y,x]\n",
      "    return s / n\n",
      "\n",
      "def compute_local_features(\n",
      "    numpy.ndarray[FLOAT32_DTYPE_t, ndim=2] img,\n",
      "    numpy.ndarray[FLOAT32_DTYPE_t, ndim=3] filter_responses_padded,\n",
      "    int num_thetas,\n",
      "    int num_samples_height,\n",
      "    int num_samples_width,\n",
      "    int patch_size):\n",
      "\n",
      "    cdef int   num_tiles_per_dimension_per_patch\n",
      "    cdef float local_feature_norm_accept_threshold\n",
      "    \n",
      "    num_tiles_per_dimension_per_patch   = 4\n",
      "    local_feature_norm_accept_threshold = 0.01\n",
      "\n",
      "    cdef numpy.ndarray[INT32_DTYPE_t,   ndim=1] samples_y\n",
      "    cdef numpy.ndarray[INT32_DTYPE_t,   ndim=1] samples_x\n",
      "    cdef numpy.ndarray[INT32_DTYPE_t,   ndim=1] tile_bounds\n",
      "    cdef numpy.ndarray[FLOAT32_DTYPE_t, ndim=2] patch\n",
      "    cdef numpy.ndarray[FLOAT32_DTYPE_t, ndim=2] tile\n",
      "    cdef numpy.ndarray[FLOAT32_DTYPE_t, ndim=1] tile_1d\n",
      "    cdef numpy.ndarray[FLOAT32_DTYPE_t, ndim=1] local_features_1d\n",
      "    \n",
      "    cdef int sample_y, sample_x, theta_index, tile_index_y, tile_index_x, patch_index_y, patch_index_x\n",
      "\n",
      "    samples_y   = numpy.linspace(0, img.shape[0]-1, num_samples_height+2).astype(numpy.int32)\n",
      "    samples_x   = numpy.linspace(0, img.shape[1]-1, num_samples_width+2).astype(numpy.int32)\n",
      "    tile_bounds = numpy.linspace(0, patch_size, num_tiles_per_dimension_per_patch+1).astype(numpy.int32)\n",
      "\n",
      "    all_local_features_list = []\n",
      "\n",
      "    for sample_y in samples_y:\n",
      "        for sample_x in samples_x:\n",
      "\n",
      "            local_features_list = []\n",
      "            \n",
      "            for theta_index in range(num_thetas):\n",
      "            \n",
      "                filter_response_padded = filter_responses_padded[theta_index,:,:]\n",
      "                patch                  = filter_response_padded[sample_y:sample_y+patch_size, sample_x:sample_x+patch_size]\n",
      "                tile                   = numpy.zeros((num_tiles_per_dimension_per_patch, num_tiles_per_dimension_per_patch), dtype=numpy.float32)\n",
      "            \n",
      "                for tile_index_y in range(num_tiles_per_dimension_per_patch):\n",
      "                    for tile_index_x in range(num_tiles_per_dimension_per_patch):\n",
      "                        tile[tile_index_y, tile_index_x] = _mean(patch[tile_bounds[tile_index_y]:tile_bounds[tile_index_y+1], \\\n",
      "                                                                       tile_bounds[tile_index_x]:tile_bounds[tile_index_x+1]])\n",
      "                        \n",
      "                tile_1d = tile.ravel()\n",
      "                local_features_list.append(tile_1d)\n",
      "                \n",
      "            local_features_1d = numpy.array(local_features_list).ravel()\n",
      "\n",
      "            if numpy.linalg.norm(local_features_1d) > local_feature_norm_accept_threshold:\n",
      "                all_local_features_list.append(local_features_1d)\n",
      "    \n",
      "    return numpy.array(all_local_features_list, dtype=numpy.float32)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 323
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "local_features = compute_local_features(img, filter_responses_padded, num_thetas, num_samples_height, num_samples_width, patch_size)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 326
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def compute_global_features():\n",
      "    local_features_codebook_indices, distances = scipy.cluster.vq.vq(local_features, local_feature_cluster_centroids)\n",
      "    hist, bin_edges                            = numpy.histogram(local_features_codebook_indices, bins=codebook_size)\n",
      "    tf_idf_hist                                = numpy.array(tf_idf_transformer.transform(hist).todense()).squeeze()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 327
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "compute_global_features()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 329
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "plot(tf_idf_hist);"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAQwAAAD9CAYAAABJNUmbAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nrkJggg==\n"
      }
     ],
     "prompt_number": 332
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def compute_distances():\n",
      "    cos_distances  = dot(tf_idf_all_global_features_2d, tf_idf_hist)\n",
      "    distances      = arccos(cos_distances)\n",
      "    sorted_indices = argsort(distances)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 333
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "print \"%s (distance = %f) \" % (img_names_exts[sorted_indices[0]], distances[sorted_indices[0]])\n",
      "print \"%s (distance = %f) \" % (img_names_exts[sorted_indices[1]], distances[sorted_indices[1]])\n",
      "print \"%s (distance = %f) \" % (img_names_exts[sorted_indices[2]], distances[sorted_indices[2]])\n",
      "print \"%s (distance = %f) \" % (img_names_exts[sorted_indices[3]], distances[sorted_indices[3]])\n",
      "print \"%s (distance = %f) \" % (img_names_exts[sorted_indices[4]], distances[sorted_indices[4]])\n",
      "\n",
      "plot(distances);"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "latitude=07.longitude=04.png (distance = 0.000000) \n",
        "latitude=08.longitude=04.png (distance = 0.405799) \n",
        "latitude=09.longitude=04.png (distance = 0.451262) \n",
        "latitude=07.longitude=05.png (distance = 0.583955) \n",
        "latitude=09.longitude=05.png (distance = 0.627951) \n"
       ]
      },
      {
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAQkAAAD9CAYAAACvHILfAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nbrDE/AAAAABJRU5ErkJggg==\n"
      }
     ],
     "prompt_number": 334
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def find_matches():\n",
      "\n",
      "    compute_filter_responses_padded()\n",
      "    local_features = compute_local_features(img, filter_responses_padded, num_thetas, num_samples_height, num_samples_width, patch_size)\n",
      "    compute_global_features()\n",
      "    compute_distances()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 335
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "%timeit -n2 -r3 find_matches()"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "2 loops, best of 3: 2.15 s per loop\n"
       ]
      }
     ],
     "prompt_number": 336
    }
   ],
   "metadata": {}
  }
 ]
}