increased time limit in nearest, also added check on shapes in AE

assignment_4_5_indexing_template/image_search.py

@@ -38,8 +38,8 @@ def nearest(means: np.ndarray, data: np.ndarray) -> (np.ndarray, np.ndarray):
 def create_db(imgs_visual_words: np.ndarray, num_words: int) -> csr_matrix:
     """
     Create database [num_words, num_imgs] of word weights represented as csr_matrix. Details explained at tutorial page.
-    imgs_visual_words is of dimensions [num_imgs, visual_words_in_img]. Number of visual_words_in_img differs for
-    each img.
+    imgs_visual_words is an array of arrays of length num_imgs. Each np.array is a list of visual words in an image.
+    Number of visual_words_in_img differs for each img.
 
     :param imgs_visual_words:
     :param num_words:
@@ -71,8 +71,8 @@ def create_db_tfidf(imgs_visual_words: np.ndarray, num_words: int, idf: np.ndarr
 def get_idf(imgs_visual_words: np.ndarray, num_words: int) -> np.ndarray:
     """
     Create Inverse Document Frequency of shape: [num_words, num_imgs]
-    imgs_visual_words is of dimensions [num_imgs, visual_words_in_img]. Number of visual_words_in_img differs for
-    each img.
+    imgs_visual_words is an array of arrays of length num_imgs. Each np.array is a list of visual words in an image.
+    Number of visual_words_in_img differs for each img.
     idf is of shape: [num_words, 1]
 
     :param imgs_visual_words:

assignment_4_5_indexing_template/spatial_verification.ipynb

+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {
+    "collapsed": true
+   },
+   "outputs": [],
+   "source": [
+    "# TODO:\n",
+    "\"\"\"\n",
+    "corrs=corrm2m(qvw, vw, relevant, max_tc, max_MxN)\n",
+    "qvw - visual words in query image\n",
+    "vw - K x visual words\n",
+    "relevant - list of relevant img indices (of len K)\n",
+    "\n",
+    "return a list of lists of tentative correspondences (K x Tk), K - #imgs, Tk - tentative\n",
+    "corrs between query and k-th relevant image (qvw_i, vw_i)\n",
+    "\n",
+    "\"\"\"\n",
+    "\n",
+    "\n",
+    "\n",
+    "# TODO: what is a minimum number of features in common\n",
+    "# M x N - cartesian product of the same visual word instances\n",
+    "# maximum min(N,M) TRUE correspondences\n",
+    "\n",
+    "# sort common visual words w.r.t. MxN\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "outputs": [],
+   "source": [
+    "from image_search import *\n",
+    "import scipy.io"
+   ],
+   "metadata": {
+    "collapsed": false,
+    "pycharm": {
+     "name": "#%%\n"
+    }
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "outputs": [
+    {
+     "traceback": [
+      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
+      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
+      "\u001b[0;32m<ipython-input-1-30f4565421d9>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mdata\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mscipy\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mio\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mloadmat\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m'data/mpvdb_haff2.mat'\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'data'\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mT\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      2\u001b[0m \u001b[0;31m# idf = getidf(VW, 10000);\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[0;31m# DB  = createdb_tfidf(VW, 10000, idf);\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0;31m#\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[0;31m# opt = [];\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
+      "\u001b[0;31mNameError\u001b[0m: name 'scipy' is not defined"
+     ],
+     "ename": "NameError",
+     "evalue": "name 'scipy' is not defined",
+     "output_type": "error"
+    }
+   ],
+   "source": [
+    "data = scipy.io.loadmat('data/mpvdb_haff2.mat')['data'].T\n",
+    "# idf = getidf(VW, 10000);\n",
+    "# DB  = createdb_tfidf(VW, 10000, idf);\n",
+    "# \n",
+    "# opt = [];\n",
+    "# opt.max_tc=600;\n",
+    "# opt.max_MxN=10;\n",
+    "# opt.max_spatial=50;\n",
+    "# % directory where you unzipped mpvimgs.zip\n",
+    "# opt.root_dir='./';\n",
+    "# % opt.subplot=1;\n",
+    "# opt.threshold=8;"
+   ],
+   "metadata": {
+    "collapsed": false,
+    "pycharm": {
+     "name": "#%% load data and create inverted file\n",
+     "is_executing": false
+    }
+   }
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "outputs": [],
+   "source": [
+    "\n"
+   ],
+   "metadata": {
+    "collapsed": false,
+    "pycharm": {
+     "name": "#%%\n"
+    }
+   }
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 2
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython2",
+   "version": "2.7.6"
+  },
+  "pycharm": {
+   "stem_cell": {
+    "cell_type": "raw",
+    "source": [],
+    "metadata": {
+     "collapsed": false
+    }
+   }
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 0
+}
\ No newline at end of file

assignment_4_5_indexing_template/spatila_verification.py

+import numpy as np
+from scipy.sparse import csr_matrix
+
+
+def get_tentative_correspondencies(
+        query_vw: np.ndarray, vw: np.ndarray, relevant_idxs: np.ndarray, max_tc: int, max_MxN: int) -> np.ndarray:
+    """
+    Compute tentative correspondencies.
+
+    :param query_vw: [num_keypoints_in_query x 1]
+    :param vw: [num_imgs x num_words_in_img] visual_words
+    :param relevant_idxs: [num_imgs, ]
+    :param max_tc: maximum tentative correspondences
+    :param max_MxN: maximum pairs
+    :return: correspondencies np.ndarray of shape [num_correspondencies x 2]
+    """
+
+    correspondencies = np.zeros((0, 2), dtype=np.int)
+
+    return correspondencies
+
+
+def get_A_matrix_from_geom(geom: np.ndarray):
+    """
+    returns a matrix
+    [
+        a11 a12 x
+        a21 a22 y
+        0   0   1
+    ]
+
+    :param geom:
+    :return: 3x3 matrix
+    """
+    return np.array([
+        [geom[2], geom[3], geom[0]],
+        [geom[4], geom[5], geom[1]],
+        [0, 0, 1],
+    ])
+
+
+def ransac_affine(
+        q_geom: np.ndarray, geometries: np.ndarray, correspondencies: np.ndarray, relevant: np.ndarray,
+        inlier_threshold: float) -> (np.ndarray, np.ndarray):
+    """
+
+    :param q_geom: query geometry of shape [num_keypoints x 6]
+    :param geometries: np.ndarray of object of len num_images. each object is np.ndarray [num_keypoints_in_img x 6]
+    :param correspondencies: [num_correspondencies x 2]
+    :param relevant: relevant indices
+    :param inlier_threshold: maximum transformed point to corresponding point distance to be considered as an inlier
+    :return: (As, inliers_counts), where As are estimated affine matrices for each image and number of inliers in each image
+    """
+
+    K = len(relevant)
+    As = np.zeros((K, 3, 3))
+    inliers_counts = np.zeros((K, ))
+
+    return As, inliers_counts
+
+
+def query_spatial_verification(
+        query_visual_words: np.ndarray, query_geometry: np.ndarray, bbox: list, visual_words: np.ndarray,
+        geometries: np.ndarray, db: csr_matrix, idf: np.ndarray, params: dict) -> (np.ndarray, np.ndarray, np.ndarray):
+    """
+
+    :param query_visual_words: [num_keypoints_in_query x 1]
+    :param query_geometry: query geometry of shape [num_keypoints x 6]
+    :param bbox: list [x, y, xx, yy]
+    :param visual_words: [num_imgs x num_words_in_img] visual_words
+    :param geometries: np.ndarray of object of len num_images. each object is np.ndarray [num_keypoints_in_img x 6]
+    :param db: [num_words, num_imgs]
+    :param idf: Inverse Document Frequency. Shape: [num_words, 1]
+    :param params: dictionary, important keys here: minimum_score, max_tc, max_MxN, use_query_expansion
+
+    :return:
+    """
+
+    num_relevant = ...  # you need to get number based on scores and params['minimum_score']
+    inliers_counts = np.zeros((num_relevant, ), dtype=np.int)
+    relevant_idxs = np.zeros((num_relevant, ), dtype=np.int)
+    As = np.zeros((num_relevant, 3, 3))
+
+    return inliers_counts, relevant_idxs, As
+